Chapter  112:  The Use of Episiotomy in Obstetrical Care: A Systematic Review

Key Questions

The original Scope of Work for this review was developed by the American College of Obstetricians and Gynecologists (ACOG) and forwarded by the Agency for Healthcare Research and Quality (AHRQ) to the RTI International-University of North Carolina Evidence-based Practice Center (RTI-UNC EPC). The work assignment proposed four provisional questions for review. Those questions were the basis for a brief review completed by the EPC Coordinating Center (The Lewin Group). Brief reviews help prioritize the topics AHRQ assigns to the 10 “generalist” EPCs.

The RTI-UNC EPC further revised the proposed questions after discussions with internal technical staff, AHRQ staff, and our Technical Expert Advisory Group (TEAG). The final key questions (KQ s) are listed below.

KQ 1. Does the practice of liberal or routine episiotomy compared to more selective use of episiotomy influence maternal postpartum outcomes?

KQ 2. Does episiotomy incision type (i.e., midline or mediolateral) influence maternal postpartum outcomes?

KQ 3. Does the repair of the perineal defect (i.e., suture type and repair approach) influence maternal postpartum outcomes?

KQ 4. Does episiotomy have a long-term influence on urinary incontinence, fecal incontinence, or pelvic-floor defects?

KQ 5. Does episiotomy or incision type, or both, influence future sexual function?

Organization of This Evidence Report

Chapter 2 describes our methods, including our search strategies and inclusion/exclusion criteria; we also document our approach to grading the quality of articles and rating the strength of evidence. In Chapter 3, we present the results of our literature search and synthesis of retained articles by key question. Chapter 4 further discusses the findings, presents our conclusions, and offers recommendations for future research. Our references and included studies and a listing of excluded studies follow Chapter 4. Appendixes include a detailed description of our search strings (Appendix A), abstraction and quality-rating forms (Appendix B), detailed evidence tables (Appendix C), and acknowledgments (Appendix D). Appendixes and evidence tables cited in this report are provided electronically at http://www.ahrq.gov/clinic/ tp/epistp.htm.

Technical Expert Advisory Group (TEAG)

We identified technical experts in the field of episiotomy to provide assistance throughout the project. The TEAG (see Appendix D) was expected to contribute to AHRQ's broader goals of (1) creating and maintaining science partnerships as well as public-private partnerships and (2) meeting the needs of an array of potential customers and users of its products. Thus, the TEAG was both an additional resource and a sounding board during the project. The TEAG included eight members: seven technical/clinical experts and one potential user of the final evidence report, an ACOG representative.

To ensure robust, scientifically relevant work, we called on the TEAG to provide reactions to work in progress and advice on substantive issues or possibly overlooked areas of research. TEAG members participated in conference calls and discussions through e-mail to

• refine the analytic framework and key questions at the beginning of the project;

• discuss the preliminary assessment of the literature, including inclusion/exclusion criteria; and

• provide input on the information and categories included in evidence tables.

Because of their extensive knowledge of the literature, including numerous articles authored by TEAG members themselves, and their active involvement in professional societies and as practitioners in the field, we also asked TEAG members to participate in the external peer review of the draft report.

Inclusion and Exclusion Criteria

Our inclusion and exclusion criteria, documented in Table 1, were relatively complex. The reason is largely that criteria for study design differed for each key question based on the sufficiency and quality of evidence. Key Questions 1 and 3 have been more commonly examined in randomized controlled trials (RCTs); thus, we elected to limit the searches to RCTs. Key Questions 2, 4, and 5 have been studied less extensively in trials; therefore, we searched for both RCTs and prospective cohort studies.

We excluded studies that (1) did not report on women of reproductive age; (2) were published in languages other than English; (3) did not report information pertinent to the key clinical questions; (4) had fewer than 40 subjects; and (5) were not original studies (although we did include systematic reviews and meta-analysis in our discussion).

Literature Search and Retrieval Process

Databases. We used multifaceted search strategies to include all the current valid research on the key questions. We used standard electronic databases: MEDLINE^®, Cochrane Collaboration resources, and the Cumulative Index to Nursing and Allied Health Literature (CINAHL^®). We also undertook hand-searches of the reference lists of relevant articles to make sure that we were not missing any relevant studies. We consulted with the Technical Expert Advisory Group (TEAG) about any studies or trials that are currently under way that may not be published yet.

Search Terms. Based on the inclusion/exclusion criteria above, we generated a list of Medical Subject Heading (MeSH) search terms (Tables 2 and 3, also Appendix A ^*). Our TEAG also reviewed these terms to ensure that we were not missing any critical areas, and this list represents our collective decisions as to the MeSH terms used for all searches. MEDLINE^® searches for “Episiotomy” articles are fairly straightforward because the concept is well established and the MeSH indexing is standard. In addition to searching on the MeSH term “Episiotomy,” we also searched for “Labor Stage, Second.”

Figure 2 presents the yield and results from our search. We conducted our initial search in late 2003 and updated it in November 2004. Beginning with a yield of 992 articles, we retained 45 articles that we determined were relevant to address our key questions and met our inclusion/exclusion criteria.

Article Selection Process. Once we had identified articles through the electronic database search, review articles, and bibliographies, we examined abstracts of articles to determine whether studies did, in fact, meet our criteria. Two reviewers separately evaluated the abstracts for inclusion or exclusion, using an Abstract Review Form (see Appendix B). If one abstractor concluded that the article should be included in the review, we retained it. Abstracts initially excluded from the study by one reviewer received a second review. The group included three physician health-services researchers—Katherine Hartmann, MD, PhD (Scientific Director); John Thorp, Jr., MD (Co-Investigator); and Gerald Gartlehner, MD, MPH (Study Coordinator); one health-services researcher—Meera Viswanathan, PhD (Study Director); and one junior epidemiologist—Rachel Palmieri, B.S.

Approximately 325 articles required review of the full article because of missing or uninformative abstracts. For the full article review, one reviewer read each article and decided whether it met our inclusion criteria, using a Full Text Inclusion/Exclusion Form (see Appendix B ^*). A list of articles excluded at the full-article review stage is provided at the end of this report, along with the reasons for their exclusion.

Development of Evidence Tables and Data Abstraction Process

The five staff members who conducted this systematic review jointly developed the data abstraction tables (see Appendix B) and evidence tables (Appendix C). These tables were designed to provide sufficient information to enable readers to understand the studies and to determine their quality; we gave particular emphasis to essential information related to our key questions. The format of the evidence tables, which was based on successful designs used for prior systematic reviews, varies somewhat by key questions.

The abstractors trained themselves on entering data into the tables by abstracting several articles and then reconvening as a group to discuss the utility of the table design. The abstractors repeated this process through several iterations until they decided that the tables included the appropriate categories for gathering the information contained in the articles. The design was then reviewed by the TEAG through a teleconference.

All team members shared the task of initially entering information into the data abstraction forms. Another member of the team also reviewed the articles and edited all initial table entries for accuracy, completeness, and consistency. The two abstractors reconciled all disagreements concerning the information reported in the abstraction forms. The full research team met regularly during the article abstraction period and discussed global issues related to the data abstraction process.

We then entered the data from the abstraction forms into evidence tables and once again checked for consistency and accuracy.

The final evidence tables are presented in their entirety in Appendix C. Entries in the tables are listed by publication date. A list of abbreviations used in the tables appears at the beginning of that appendix.

Quality and Strength of Evidence Evaluation

Rating the Quality of Individual Articles. The RTI-UNC EPC's approach to assessing the quality of individual articles was developed based on the domains and elements recommended in the evidence report by West and colleagues, Systems to Rate the Strength of Scientific Evidence. ²⁶ We developed different rating schemes for RCTs and prospective cohort studies.

For RCTs, we rated studies on the following criteria (see Appendix B for the RCT Quality Rating Form).

• 1. Randomization Approach and Implementation: This item judged whether the approach described a valid method of randomization, whether allocation concealment was achieved, and whether balance was documented across study groups.

  Approach: Articles that assigned the groups in a manner inconsistent with true randomization methods automatically received a poor rating for this category and overall. Articles that merely stated that they “randomly assigned” the groups and had either no balance or did not report on balance received a poor rating. Articles with no documentation of concealment were rated poor. Those with potentially inadequate concealment methods were rated poor if the study had poor balance of allocation or if balance was not documented in the paper. Those with potentially poor concealment were rated fair if they documented good balance.

• 2. Masking: This item was relevant only to Key Question (KQ) 3. For KQ 1, masking of the birth attendant is not feasible and for KQ 2, masking the kind of episiotomy the women received was not possible.

  Approach: If the outcome assessors and participants were adequately masked within the possibilities of the study design, we rated the category as good. If there was a mix of masking among the outcomes, we rated the category as fair. If masking was not done at all and not attempted, we rated the category as poor. In the event that the article did not report on masking, we noted this point in the quality assessment table and counted it against the trial in the overall quality rating.

• 3. Operational Definitions and Measurements: This item judged the quality of the operational definitions of the outcomes (i.e., were they adequately described) and whether they were adequately collected (i.e., was the method sufficient and appropriate).

  Approach: If a primary outcome was identified, we gave it more weight in its contribution to this category's score. Otherwise, we rated this category on the basis of an average across all outcomes and the ability to define and measure them. Good definitions and measurement include the following: visual analog scale, detailed Likert scale, detailed time points in question, details about what was asked of the patient, medical chart abstractions, and clinical examination or assessment. If an article simply stated an outcome, such as “perineal pain,” and gave no further explanations about it, we rated the category in the fair-to-poor range, depending on how the study collected the information.

• 4. Post-Randomization Exclusions: This item captured how many post-randomization exclusions were explicitly stated.

  Approach: In typical randomized trials, intention to treat analysis is expected. Some investigators represented in this literature enrolled women during prenatal care rather than on labor and delivery in an effort to get a representative sample of prenatal patients. We note exclusions as appropriate when individual gave birth at a hospital not participating in the study, or when participants had outcomes that made them ineligible to participate in the trial such as preterm birth or cesarean birth. Any other exclusions after randomization were considered inappropriate.

• 5. Loss to Followup: This item collected percentages of followup at every time point in the study at which data were collected; we used it to determine if followup was adequate.

  Approach: An average of followup percentages for short-term and long-term followup contributed to this category. In general, we considered followup greater than or equal to 90 percent in the short term and 80 percent in the long term to be good.

• 6. Statistical analysis: This factor included whether the investigators conducted the study in an appropriate manner and took the effect of multiple comparisons into account. This item also reviewed the study's use of multivariate statistical techniques and/or participant restriction or stratification to control for confounding.

  Approach: This category is not included on the quality assessment form because of the nature of reporting in journals dating back to 1974. P-values were sufficient for reporting in the past, whereas point estimates, tests for homogeneity, stratification, and confidence intervals are more widely reported now. Although this category did not explicitly contribute to the overall quality rating, we used it for articles that were on the border between categories.

For RCTs, the two article abstractors independently rated each article on each of the first five categories as indicated by the quality assessment form (Appendix B ^*). A third reviewer flagged studies with differences in scoring on individual components. We reconciled these differences by consensus. We then created a composite rating. If a study had poor randomization approach or implementation with a fatal flaw (e.g., lottery cards), we rated it as poor. For all other scores, we gave each item equal weight. Specifically, studies that received good ratings on all categories were rated as good studies overall. If a study received one or two fair or poor ratings, or the equivalent of a deficiency, it was rated as an overall fair-quality study. Studies with three or more fair ratings or a poor randomization design or implementation with a fatal flaw were rated poor-quality studies.

For classifying the quality of prospective cohort studies included for KQs4 and 5, we assessed the following factors:

• 1. Study population: We sought documentation in the publication of the degree to which the study population was representative of women with uncomplicated spontaneous vaginal births in the study facilities or broader population sampled.

• Approach: To receive a rating of good for this component of study design and conduct, we required a study to describe clearly (1) the base population from which cohort participants were sought, (2) the number of women in that base population (a denominator), (3) clear inclusion and exclusion criteria, and (4) the proportion of eligible women who were ultimately enrolled in the cohort.

• Studies lacking only items (2) and (4) were classified as fair, and studies lacking items (1) or (3) with any combination of other missing documentation were rated as poor with respect to documentation of the study population.

• 2. Measures: We sought documentation in the publication of four components of quality of measurements. The first was specification of whether the measure was a primary or secondary measure for the study as noted in any portion of the paper. The second was a clear description of the measures used that is sufficient to allow replication of the measure (e.g., visual analog scale, McGill Pain Score). We accepted references to methods described more fully in other publications as documentation if the reference in fact provided details. The third component was a clear description of how the measure was obtained and by whom if, applicable (e.g., telephone interview, face-to-face interview, mailed questionnaire). The fourth was a clear specification of the time interval in which the data were collected with respect to the index birth.

• Approach: We classified studies that achieved all four document requirements for the measurement of relevance to the key question as having good implementation of the measures component. We classified studies as fair for this component if item (1) was unclear or not noted and if this was the only limitation. If any other item was missing, we considered the quality of documentation of measures to be poor. Of note, a given study could be classified as good for one key question (e.g., the key question about sexual function), while getting a fair rating for another measure that related to a different key question.

• 3. Loss to followup: If data from more than one time interval were reported, we sought documentation of the these followup measures: (1) the number of participants in the sample at the time of followup, (2) analysis of how respondents differed from nonrespondents if loss exceeded 20 percent, and (3) absolute loss to followup by time interval.

• Approach: We rated a study as good quality if the research team accomplished each of the above measures and had ≤ 20 percent loss to followup at 3 months and beyond. A study was rated fair if the investigators accomplished items (1) and (2), had no apparent response bias as investigated by comparison of baseline characteristics, and had up to 30 percent loss to followup; or if they had between 20 percent and 25 percent loss to followup without documentation of comparability. We rated a study as poor for this component if it had more than 30 percent loss to followup or more than 25 percent loss without comparison for response bias.

• 4. Analysis: We sought four tiers of documentation: (1) thorough enumeration of the number of cohort participants, the characteristics of their birth experience and perineal status, and general descriptive characteristics such as parity and number of prior vaginal births in cohorts that included multiparous women; (2) assessment of confounding and modifying factors by bivariate analysis, stratified analysis, or multivariable modeling; (3) reporting of adjusted estimates for main effects that took into account identified confounding or modifying factors (stratified or separate analyses were acceptable for simple constructs); and (4) presentation of adjusted results with a measure of statistical precision such as a confidence interval or P-value.

• Approach: We rated a study as having a good analysis implementation if all of these elements were present. Missing or limited detail for item (1) resulted in a fair rating if this was the only deficit; similarly, we rated a study fair if it was missing or providing only limited detail for item (2) if subsequent multivariable modeling implied that the step had been completed and all other items were present. Missing items (3) or (4) or any other two or more items in combination resulted in a poor rating.

Grading the Strength of Available Evidence. Our scheme follows the criteria applied by Berkman et al.²⁷ That system included three domains: quality of the research, quantity of studies (including number of studies and adequacy of the sample size), and consistency of findings. The four senior staff members assigned grades by consensus.

We graded the body of literature applicable to each of the four components of the two key questions separately and present our findings in Chapter 4. The possible grades in our scheme are as follows:

• I. The evidence is from studies of strong design; results are both clinically important and consistent with minor exceptions at most; results are free from serious doubts about generalizability, bias, or flaws in research design. Studies with negative results have sufficiently large samples to have adequate statistical power.

• II. The evidence is from studies of strong design, but some uncertainty remains because of inconsistencies or concern about generalizability, bias, research design flaws, or adequate sample size. Alternatively, the evidence is consistent but derives from studies of weaker design.

• III. The evidence is from a limited number of studies of weaker design. Studies with strong design either have not been done or are inconclusive.

• IV. No published literature.

Literature Search and Included Studies

Overview of the Evidence. We identified seven primary publications of randomized controlled trials (RCTs) of liberal versus restrictive use of episiotomy.^{22, 23, 28–32} Evidence Table 1 in Appendix C provides details on each study. Three of these studies were conducted in the United Kingdom;^{22, 23, 28} one at a British military hospital in Saudi Arabia;³¹ one in Germany;³² one in Argentina;³⁰ and one in Canada.²⁹ The first trial was performed in the United Kingdom and published in 1984;²³ the most recent, in Germany in 2004.³² The studies are not evenly distributed over time: three trials were conducted in the early to mid-1980s in the United Kingdom, three in the first half of the 1990s, and a decade passed before the publication by Danneker and colleagues in Germany in 2004. We also identified two secondary reports from randomized clinical trials.^{33, 34} One publication reports on 3-year followup of a trial cohort;³³ the other is a re-analysis of 3-month followup data grouped by degree of perineal trauma after the birth, rather than by allocated trial group.³⁴ Discussion of these publications is included in the results for Key Questions 4 and 5 because they focus on continence and sexual function outcomes.

Each trial that we identified, compared two study arms or groups: (1) one in which the intention was to restrict routine use of episiotomy and (2) one group in which a liberal--use policy endorsed routine use. Defining the distinctions between such groups in a manner that can be uniformly achieved is the central challenge for these trials.

The strictest definition of “restrictive” was to avoid episiotomy unless indicated for fetal well-being.^{23, 32} Other definitions pivoted on instructions to “avoid episiotomy,” use only when “medically necessary,” or not perform episiotomy for the sole purpose of avoiding a laceration.^{22, 28, 29, 31} The largest trial defined restrictive use as only for fetal indications and/or to avoid severe lacerations.³⁰

Liberal-use arms were defined in terms such as “routinely conducted,” “routine,” “usual care,” and “elective.”^{22, 28–31} Two studies describeed the liberal-use policy as encouraging routine use of episiotomy when “a tear is imminent”³² and “to prevent a tear.”²³

Regardless of how the randomization groups were defined for study implementation, neither definition may align with the usual practice of individual clinicians, especially with respect to how they would describe the goal they are trying to achieve when performing an episiotomy. Variation in norms and usual practice patterns are demonstrable in the variation of episiotomy use observed in these trials. Use of episiotomy in the restrictive groups ranged from lows of 7.6 percent²² and 10.2 percent²³ to highs of 44 percent²⁹ and 53 percent.³¹ We observed up to a seven-fold difference in the use of episiotomy within the restrictive groups across these trials. The routine--use/liberal-use arms had episiotomy use rates from a low of 44.9 percent²² to a high of 83 percent.^{30, 31} Wide variation in patterns of use across trials introduces substantial heterogeneity in the “exposures” under study. A large degree of built-in “cross-over” occurred in the setting of higher rates of use in the restrictive groups. In synthesizing these data, we emphasize that these trials compare policies of episiotomy use, not episiotomy versus no episiotomy.

An additional factor that influences generalizability of findings for practice in the United States is that six of the seven studies used mediolateral episiotomy. The only North American study, conducted in Canada, was also the only study in which median (midline) episiotomy was used. Because midline episiotomy is the most common technique used in the United States,^{9, 35} this means that the majority of the literature reflects outcomes that would be expected with a distinctively different episiotomy approach with respect to anatomic location of the defect and potential complications. Key Question 2, focused on incision type, identified only one poor- quality RCT directly comparing median and mediolateral episiotomy.³⁶ We review this study- in detail in the next section; briefly, the study -suggests increased risk of rectal injury and complicated or extended incision with midline episiotomy. This finding is comparable to those of observational studies that report that midline incisions are more likely to result in extensions that injure the anal sphincter and/or rectal mucosa.^{37, 38}

Study Populations. Six studies restricted participation to term births;^{22, 23, 28–31} the seventh, to births at longer than 34 weeks' gestation.³² Five studies specified that they enrolled only singleton gestations. The two studies that did not specify singleton gestations were conducted in the 1980s before routine use of ultrasound; in that period, -providers were unlikely to miss a multiple gestation clinically, -excluding twin gestations before enrollment- in a trial might have been difficult to do with complete confidence.^{22, 23} Two studies required vertex presentations.^{28, 31} Regardless of stated inclusion criteria, multiple gestations and breech presentations do not seem to be represented in these trials. To this extent the studies do represent episiotomy use in uncomplicated vaginal deliveries.

Three studies enrolled only women having their first births.^{22, 31, 32} This approach eliminates any influence of prior perineal trauma and healing on the trial outcomes. The combined study populations of these three studies was 409 participants; this is smaller than the entire cohort of each of the other studies. Four studies enrolled multiparous women. In these studies, the proportion of women who were primiparous ranged from 40 percent to 68 percent, generally with good balance across study groups.^{23, 28–30} The exception occurred in the publication by House and colleagues in which it appears that multiparous women were more likely to have been randomized to the restricted-use group.²⁸ Within the studies that achieved balanced allocation by parity, analyses that stratify outcomes by parity help inform how outcomes may differ based on prior childbirth experiences. No trial data were identified that allow consideration of whether outcomes vary by race and ethnicity.

Each study focused on normal spontaneous vaginal births. To reduce the number of women who subsequently had operative vaginal deliveries or cesarean births, the majority of studies allocated women to study groups as close to the anticipated time of birth as feasible. The proportion of assisted vaginal births (both forceps and vacuum) in these trials was 2 percent,³⁰ 3 percent,²⁹ 4 percent to 5 percent,³¹ 11 percent to 14 percent,²⁸ or not specifically reported; . The absence of reporting on instrumental and cesarean births raises the potential of unreported post-randomization exclusions. In two cases, authors noted the number of cesarean births and that those cases are described as part of the trial population and excluded from analyses.^{29, 32} Both of these studies enrolled women during prenatal care, an approach that improved representativeness of the study population and explains the increased numbers of women with cesarean births who were logically excluded later.

Outcomes. The most common primary outcome was perineal status after the birth. All seven studies reported incidence of episiotomy and of third- or fourth-degree lacerations or extensions. Five studies reported prevalence of intact perineum;^{22, 23, 29, 31, 32} one reported prevalence of intact perineum combined with first-degree lacerations.²⁸ One trial did not report data about intact perineum or minor lacerations, but the proportion can be inferred because first-degree or intact should be the converse of the data they do report about “any perineal suturing” required.³⁰

A single study incorporated masked assessment of perineal trauma; Sleep and colleagues arranged to have a clinician who did not assist the delivery and was masked to the study group asesss perineal status and perform the repair.²³ Although this design approach does not mask the obvious appearance of an episiotomy compared to a spontaneous tear or intact perineum, it prevents any bias in uniformly recording the extent and location of any perineal trauma. The other studies are at risk of this type of bias.

The most common secondary outcome was pain in the days immediately after the birth. Five of the seven trials assessed pain. Two groups used visual analog scores and classified responses into categories of mild, moderate, or severe.^{28, 32} One study used an unspecified “standardized questionnaire” and also reported pain severity as mild, moderate, or severe. ²³ Another used the McGill Pain Questionnaire and reported the composite score of the 10-item scale.²⁹ The largest study, conducted in Argentina, did not define how they collected data about “perineal pain.”³⁰ Additional measures include use of pain medications and reports of pain with specific physical activities.

Three publications did not report masking of the assessors to study group.^{22, 31, 32} The remainder of the studies reported that the individual conducting the pain assessment was unaware of allocation of the participants. Two studies clearly noted that women were not aware of the group to which they were allocated.^{23, 28}

Additional outcomes assessed include resumption of sexual activity, estimated blood loss, and wound- healing appearance and complications, including infection, healing by secondary intention, and persistent granulation tissue. Few of these outcomes were measured uniformly in more than one study.

Results

Perineal Trauma Outcomes. Table 4 summarizes key elements of study design and the perineal outcomes observed in each of the RCTs. The table provides a summary and does not include all outcomes such as anterior tears. Based on quality of the conduct of the trials, the strongest of these studies (rated good quality) was the first RCT conducted; like the majority of these trials, it evaluated the use of mediolateral episiotomy. This first study, conducted in the early 1980s, enrolled 1,000 women, featured an appropriate randomization approach that achieved good balance, and had masked assessment of outcomes with clear definitions.²³ The investigators did not make any post-randomization exclusions and specifically reported conducting an intention-to-treat analysis. The study methods achieved a wide gradient of episiotomy use between the liberal and restrictive groups—a 41 percentage point difference: 51.4 percent in the liberal-use group and 10.2 percent in the restrictive- use group. Women in the restrictive group were more likely to have an intact perineum; 24.3 percent of those in the liberal group had intact perinea compared to 33.9 percent in the restrictive group. Third- and fourth-degree lacerations were rare (0.2 percent in the trial cohort) and did not differ by group. The effects of liberal versus restricted use with respect to any need for suturing were less marked among multiparous women who had similar outcomes: 69 percent of multiparous women in the liberal group and 66 percent in the restrictive group. required any suturing for repair. Among nulliparous women, the difference in need for suturing was more pronounced (89 percent in the liberal-use group and 74 percent in the restrictive group). Sleep and colleagues concluded that restricting use of episiotomy neither increased nor decreased maternal morbidity.²³

The largest trial conducted was a multisite study in Argentina that enrolled 2,606 women and was of fair quality (see evidence tables in Appendix C for details).³⁰ This study found a 2.4-fold increase in risk of anterior tears among women in the restrictive- use arm (95% confidence interval [CI], 1.9–2.9), and decreased risk of posterior perineal surgical repair (relative risk [RR] = 0.72; 95% CI, 0.68–0.75) when comparing restrictive to liberal use. Eighty-eight percent of women in the liberal group had a surgical repair, as did 63 percent in the restrictive group. Pain, healing complications, and dehiscence were all less frequent in the restrictive-use group. The authors concluded that episiotomy confers no apparent benefit and that high rates of use cannot be justified.

These two studies, which are also the two largest trials, yield results compatible with the findings of the other trials with respect to perineal outcomes (see Table 4 for summary). Intact perineum was uniformly more common in the restrictive groups.^{28, 29, 31, 32} With two exceptions,^{29, 32} studies reported more third- and fourth-degree lacerations among women in the liberal-use group. However, each of these studies was underpowered to distinguish differences in risk across the groups because third- and fourth-degree lacerations were rare. In fact, one study of 200 women had no severe lacerations among participants in either arm.³¹ Anterior lacerations, including anterior labial lacerations, were reported to be more common in three studies^{23, 30, 32} and equivalent in one study.³² Anterior lacerations did not contribute to overall higher use of suturing, suggesting that these tears were less severe than posterior tears.

The findings of these studies are fully compatible. None of the authors concluded that episiotomy provides any benefits with respect to perineal trauma. The majority concluded that intact or minimal perineal trauma is more common when episiotomy use is restricted. This synthesis of the data is compatible with the findings of prior systematic evidence reviews that are updated with this report. Although the authors of these research publication appear loath to ascribe harm to the use of episiotomy, in our judgement-concluding from their data that routine use is harmful is accurate, at least to the extent that it creates a surgical incision of greater extent than a woman might have experienced had the episiotomy not been performed.

Pain Outcomes. Five studies assessed pain outcomes (Table 5).^{23, 28–30, 32} The two largest trials collected pain outcome data.^{23, 30} Sleep and colleagues used midwives masked to study group to assess pain at 10 days postpartum. Participants were asked to assess pain severity in the prior 24 hours. Pain severity groupings were virtually identical by study group. Among those in the liberal-use group, 14.6 percent had mild pain; 7.8 percent, moderate; and 0.2 percent, severe; the comparable proportions for the restricted-use group were 14.1 percent, 7.5 percent, and 0.9 percent, respectively. Use of oral analgesics by postpartum day 10 was rare in both groups—2 percent and 3 percent, respectively—and not different by group. The Argentine study did not adequately define how they measured pain; pain is reported as “pain on the day of discharge”; the liberal-use group is reported as 42.5 percent with pain, and the restricted-use group as 30.7 percent with pain.

The most recent study, although small, provided the most nuanced approach to pain assessment. The investigators used a visual analog scale to assess pain with four activities: bedrest, sitting, walking, and defecation. Scores were reported in millimeters of the 100 mm-pain scale for each activity. Those in the liberal-use group had the following mean scores (± standard deviation): 39 ± 28 mm with bedrest; 69 ± 23 mm with sitting; 56 ± 24 mm with walking; and 36 ± 30 mm with defecation. The comparable scores in the restrictive group were 22 ± 21 mm; 51 ± 25 mm; 37 ± 24 mm; and 21 ± 21 mm. These differences indicate that the restrictive-use group experienced significantly lower perineal pain during all activities, at levels that are likely clinically significant. These findings could indicate a real difference in pain outcomes or some bias in assessment. The publication does not report masking of the assessors or how the assessments were conducted.

House and colleagues report that pain was more severe on postpartum day 3 among those in the liberal-use group. They also assessed pain outcomes by visual analog scale during an interview conducted by an author; masking of the assessors is not specifically noted. On day 3 in the liberal-use group, 55 percent of women had mild pain' 34 percent, moderate pain; and 11 percent, severe; the comparable categories for those in the restrictive-use group were 68 percent, 22 percent, and 10 percent. They also report tenderness at the time of examination on postpartum day 3. The restricted use group had less tenderness on examination: 79 percent had mild or minimal pain; 18 percent, moderate; 3 percent, severe; compared to 51 percent, 39 percent, and 10 percent in the liberal-use group. . These differences were statistically significant and likely to be clinically relevant.²⁸ These differences in pain by group had resolved by 6 weeks and 3 months respectively.

Klein and colleagues, who conducted the only North American trial and the only trial using midline episiotomy, found no difference in McGill pain scores on days 1, 2, and 10 after the birth for either perineal pain or pain with urination.²⁹ They reported that they conducted analyses both with individual pain-scale items and composite scores. Across each of the five studies, no study found a pain measure that was improved by routine liberal use of episiotomy.

Healing Outcomes. Two studies assessed healing outcomes by physical examination. The Argentine trial reported no difference in rates of the following adverse outcomes that were adequately defined: hematoma prior to discharge; - and infection, healing complications, and dehiscence as assessed on day 7 postpartum. At discharge, they assessed 92 percent and 93 percent of participants in the restricted use and liberal-use groups respectively; this dropped to 43 percent of both groups by the evaluation on day 7 postpartum. ³⁰ House and colleagues examined participants on day 3 postpartum and at 6 weeks. Risk of infection was assessed for all participants on day 3; poor wound apposition and granulation tissue indicating secondary healing were assessed at the later visit at which 53 percent of the trial participants were assessed. Each adverse outcome was equivalent across groups.²⁸

Other Outcomes. Two trials, both reflecting use of mediolateral episiotomy, reported on timing of resumption of intercourse. One study documented that women in the restricted episiotomy-use group resumed intercourse an average of a week earlier (5.5 ± 3.0 weeks compared to 6.5 ± 3.0 weeks).²⁸ The other study found that 37 percent of the women in the liberal-use group (P < 0.01) .²³ Longer-term influences on sexual function - those assessed at 3 months or later - are reviewed in the section on KQ 5.

Two studies, also of mediolateral episiotomy use, assessed estimated maternal blood loss. One found no difference in the amount that maternal hemoglobin measures fell.³² The other found that estimated blood loss (method not defined) was 58 cc greater in the liberal-use group, a statistically significant but likely not clinically relevant mean difference.²³

Literature Search and Included Studies

We found only one RCT comparing outcomes of midline episiotomy to those of mediolateral episiotomy.³⁶ Evidence Table 2 in Appendix C provides details. An additional focused literature search did not reveal any prospective cohort studies pertaining to this key question.

Study Participants. The study included primigravidas who were admitted to the delivery unit of a university hospital in London. The mean age at delivery was 26 years; the mean gestational age was 40 weeks.

Episiotomy Type. Midline episiotomy—“incisions divided 2 to 3 cm of the perineal tissue in the midline.” Mediolateral incisions “were made from the midline and were carried to the right of the anal sphincter for about 3 to 4 cm.”³⁶ Method of repair was identical for both study groups. Standard repair technique for both types of incisions included subcuticular skin closure with polygycolic acid suture.

Outcomes. Outcome measures included the proportion of extended or complicated incisions, recommencement of sexual intercourse, pain, pain during intercourse, satisfaction from intercourse, and the cosmetic appearance of the scar. Investigators did not report methods of outcomes assessment or the use of objective scales for pain and sexual satisfaction. An exam was conducted and pain was queried before hospital discharge and again at a 3-month followup visit that included the sexual function data collection. The study did not assess differences in fecal incontinence.

Quality. This study had serious methodological flaws; we gave it a poor rating for internal validity. In particular, we viewed an inadequate randomization method, lack of allocation concealment, and failure to blind the outcome assessors as potential sources of severe biases and as a rationale for a poor-quality rating. Consequently, the evidence is insufficient to draw any firm conclusions on differences in adverse outcomes of midline compared to mediolateral episiotomy.

Results

The trial included 407 primigravidas who were randomly assigned to midline or mediolateral episiotomy. Results revealed a significantly higher rate of complicated or extended incisions for the midline group than for the mediolateral group (P < 0.001). A total of 23.9 percent of women in the midline group experienced an extension of the episiotomy into or through the sphincter, compared to 9 percent of women in the mediolateral group. The midline group had significantly less bruising of the perineum than the mediolateral group (P < 0.001). The investigators did not note any differences in pain in the postpartum timeframe. Of all enrolled women, 76 percent attended a 3-month followup. Women in the midline group began sexual intercourse significantly earlier (P < 0.01) and had a significantly better cosmetic appearance of the scar (P < 0.02) than women in the mediolateral group. No significant differences in pain or satisfaction from sexual intercourse were detected.

On the question of midline versus mediolateral episiotomy, the only study in our review that met our inclusion criteria found that women receiving midline episiotomy had a significantly greater probability of anal sphincter injuries than women in the mediolateral episiotomy group.³⁶ This study did not assess fecal incontinence as a long-term health outcome. Because of considerable methodologic flaws, any conclusions must be made cautiously.

Differences in sphincter injury rates are clinically important. As an RCT, this study's internal validity is poor. Nevertheless, considering that this is the only trial pertaining to this key question, the findings regarding sphincter injuries could be viewed as relevant observational evidence. Multiple retrospective cohort studies that did not meet eligibility criteria for this key question support findings regarding high sphincter injury rates and midline episiotomy.^39–45 These studies provide consistent evidence that midline episiotomy leads to significantly higher rates of third- and fourth-degree tears than mediolateral episiotomy. In another study, women with midline episiotomy had a significantly higher rate of fecal incontinence than did women with spontaneous second-degree tears.⁴⁶

Literature Search and Included Studies

Overview. We included 17 randomized controlled trials (RCTs) examining various methods and materials for repairing perineal defects.^{47, 48, 48–66} As shown in the three main sections of Table 6, four trials investigated techniques of repair;^{48, 58, 60, 63, 66} 14 trials investigated materials for repair;^{50, 47, 48, 51–54, 56–59, 61–65}and two trials combined comparison of both techniques and materials in their design.^{49, 55} Details on all 17 studies are provided in Evidence Tables 3–10 in Appendix C.

Of the four trials investigating techniques of repair, two compared a two-layer approach (leaving the perineal skin unsutured) with a three-layer approach (suturing the perineal skin) and two trials compared a continuous (subcutaneous) technique with an interrupted (transcutaneous) technique.

Of the 14 trials investigating repair materials, eight compared polyglycolic-acid sutures with chromic-catgut sutures, both absorbable. Two trials compared absorbable sutures (one polyglycolic acid and one chromic catgut) with an enbucrilate tissue adhesive (Histoacryl^®). Two trials compared standard absorbable suture material with its rapidly absorbed counterpart, and one trial compared untreated chromic-catgut with a glycerol-treated “softgut” chromic catgut. In addition, two trials compared nonabsorbable and absorbable sutures: one compared silk sutures with polyglycolic-acid sutures and one compared silk sutures with both polyglycolic-acid and chromic-catgut sutures.

Most of these trials randomly allocated participants to one of two groups. Three trials, however, incorporated a factorial design of randomization. Using a 2×2 design, both the Ipswich Childbirth Study^{60, 61, 63} and the Kettle et al. trial⁶⁷ randomized to methods of repair and type of sutures. The Mahomed et al. perineal suture study used a 2×3×2 design and randomized to suture type for deep tissue repair (two groups), suture type for the perineal skin (three groups), and method of repair (two groups).⁵⁸ These studies contributed to more than one section of the results below.

Country. Approximately 65 percent of the RCTs in this report were conducted in the United Kingdom, including Ireland and Scotland. Australia, Denmark, Israel, Malaysia, Nigeria, and the United States each contributed one trial to this report.

Episiotomy Type. These trials included women who had an episiotomy at the time of vaginal childbirth, regardless of the number of previous pregnancies or births. The vast majority of episiotomies repaired in these studies were mediolateral. This reflects the fact that most of the studies were conducted in European countries, the majority contributed by the United Kingdom. Practitioners in North America generally perform midline episiotomy, whereas mediolateral is the rule elsewhere.

Because this question addresses outcomes of repair, we reviewed trials that included women who had forcep- or vacuum-assisted births with an episiotomy. The rationale for inclusion is that the technique of repair and materials can be evaluated with respect to postpartum perineal healing and maternal morbidity, regardless of the mechanisms that led to the perineal trauma. Of these 17 trials, seven explicitly excluded women who had instrumental deliveries; four did not report whether instrumental deliveries were included; one trial included instrumental deliveries but did not report the proportion of women who had them; and five trials reported proportions of participants who had instrumental deliveries (see Table 6). The proportion of participants who had instrumental deliveries ranged from17 percent to 64 percent.

Outcomes. Perineal pain and need for analgesia were assessed during the short-term postpartum period in a majority of the trials and during the long-term postpartum period for some of the trials. Investigators used self-report through interviews administered by midwives or study staff and questionnaires to measure subjective levels of pain and use of analgesia. Trials also reported on specific aspects of healing and wound breakdown including inflammation, bruising, infection, wound gaping, need for removal of sutures, and need for resuturing. In each case, a clinician involved in the trial assessed these outcomes, often without masking to study allocation. Longer-term outcomes related to sexual function, such as dyspareunia, recommencement of sexual intercourse, and timing of resumption of intercourse, were typically measured at 3 months postpartum and up to 3 years by interview or questionnaire. Incontinence and other pelvic-floor-related outcomes were investigated by one trial.⁵⁵ Few trials collected data on comfort with daily activities or satisfaction with repair.

Quality. Of the 17 trials, we rated three as good quality, eight as fair quality, and six as poor quality; for the last group, four trials were rated poor because of inadequate randomization techniques and were most likely not truly randomized. We rated trials as fair or poor quality on the basis of inadequate randomization approach and implementation, failure to mask the outcome assessors, and high loss to followup. Specific limitations of the trials are discussed elsewhere in this report.

Results for Methods of Repair

Two-layer vs. Three-layer Repair. One trial of good quality^{60, 63} and one trial of fair quality⁶⁶ compared a two-layer suturing approach with a three-layer suturing approach (Evidence Table 3, Appendix C). Both trials described the standard technique as three-layered, suturing the perineal skin closed after repair of the vagina and deeper tissues with interrupted transcutaneous or continuous subcuticular sutures. The two-layer approach leaves the perineal skin unsutured, a technique that is hypothesized to decrease perineal morbidity.

Pain and analgesia use. Consistent evidence from these two trials suggests that the two-layer suturing technique decreases perineal pain in both the short- and long-term postpartum periods and requires less analgesia use. When the evidence is limited to the trial of good quality, however, these differences were not statistically significant.

Although both trials reported less perineal pain in the two-layer groups, only one trial⁶⁶ found significant differences in the short-term postpartum period (0 to 3 months) and the long-term postpartum period (≥ 3 months). At 48 hours postpartum, fewer participants in the two-layer approach reported perineal pain than those in the three-layer approach (RR = 0.87; 95% CI, 0.78–0.97). Over time, the differences persisted: two-layer was superior at 14 days (RR = 0.77; 95% CI, 0.61–0.98), at 6 weeks (RR = 0.64; 95% CI, 0.44–0.93), and at 3 months postpartum (RR = 0.19; 95% CI, 0.06–0.54). The Ipswich Childbirth Study^{60, 63} did not note differences between the two groups in self-report of any, mild, moderate, or severe perineal pain at 24 to 48 hours, 10 days, 3 months, or 1 year. At each followup period, however, fewer women in the two-layer group reported perineal pain.

Similar findings extended to analgesia use in the two groups but only in the short-term postpartum period. Participants in the two-layer group of the Oboro et al. trial reported significantly less use of analgesics at 48 hours (RR = 0.71; 95% CI, 0.60–0.83) and at 14 days (RR = 0.54; 95% CI, 0.32–0.90) but not at 6 weeks (RR = 0.56; 95% CI, 0.16–0.1.89) or at 3 months postpartum (RR = 0.16; 95% CI, 0.02–1.34). The Ipswich trial found no differences in analgesia use at followup.^{60, 63}

Healing and wound breakdown. Despite inconsistent definition and measurement of healing outcomes, the evidence suggesting that the two-layer approach decreases healing complications and that wound gaping associated with leaving the perineal skin unsutured resolves shortly after repair.

Significantly fewer participants in the two-layer group needed sutures removed for pain or infection in both trials at any time. Healing outcomes were assessed at 14 days, 6 weeks, and 3 months in one trial⁶⁶ and were generally significant ; these outcomes were also significant at 1 year in the other trial (RR = 0.61, 95% CI, 0.45–0.83, P = 0.002).^{60, 63}

Oboro et al. found that fewer women in the two-layer group reported “tight stitches” than in the three-layer group.⁶⁶ Neither trial found a difference in the need for resuturing.

In the Ipswich trials, more women in the two-layer group than the three-layer group had wound gaping, defined in one of the trials as having edges greater than 0.5 cm apart, in the 24- to 48-hour postpartum period.⁶⁰ Observing such a separation may be an artifact of the technique used to examine the incision, because only incisions without suture approximation of the skin can appear to “gape” when tension is applied to the posterior perineum to allow visual inspection. Only one trial looked at outcomes at 1 year postpartum; the authors reported that fewer women in the two-layer group reported that the area that had been cut or torn felt different (RR = 0.75; 95% CI, 0.61–0.91; P < 0.01).⁶³

Incontinence and pelvic floor function. Neither trial investigated outcomes related to incontinence and pelvic floor function in terms of the difference in suturing methods.

Sexual function. These two trials did not investigate the same sexual functioning outcomes. The trend in healing outcomes suggests that the two-layer approach to suturing is associated with less morbidity.

Women in the two-layer repair group were significantly less likely in the Oboro et al. trial to have superficial dyspareunia than women in the three-layer repair group at both 6 weeks and 3 months (RR = 0.60; 95% CI, 0.42–0.85 and RR = 0.52; 95% CI, 0.33–0.81, respectively).⁶⁶ These results did not extend to deep dyspareunia, which was comparable between the two groups. At 6 weeks, more women in the two-layer group had resumed pain-free intercourse (26 percent vs. 10 percent, RR = 2.54; 95% CI, 1.82–3.55);⁶⁶ in the Ipswich trial,^{60, 63} this difference was noted but not statistically significant.

Continuous vs. Interrupted Sutures. Two trials, both of good quality, investigated perineal outcomes in women whose repairs were made with a continuous, subcutaneous suturing method compared to an interrupted, transcutaneous method^{48, 58} (Evidence Table 4, Appendix C).

Pain and analgesia use. The evidence is inconsistent as to whether the continuous subcutaneous method of suturing decreases perineal pain and need for oral analgesia following repair. In the Mahomed et al. trial,⁵⁸ the groups did not differ across the categories of any, mild, moderate, or severe perineal pain at day 2, day 10, and 3 months postpartum. By contrast, the trial by Kettle and colleagues documented significant differences.⁴⁸ In this latter trial, at the same time points, women in the continuous-suture group reported less pain than women in the interrupted-suture group. On day 2, 69 percent of the continuous group and 79 percent of the interrupted group reported pain (P < 0.0001). The difference in pain outcomes continued at day 10 (26.5 percent vs. 44 percent; P < 0.0001), 3 months (9 percent vs. 13 percent; P = 0.03) and 1 year (4 percent vs. 7 percent; P = 0.05). At day 10, significant differences were also seen in perineal pain reported during walking, sitting, urination, and defecation; the difference in favor of less pain among those with subcutaneous repair ranged from 7 percent to 16 percent.

The former study also did not find significant differences between the two groups in their need for oral analgesia at day 2 (52% vs. 48%) or day 10 (7% vs. 9%). By contrast, the latter trial found that women in the continuous group needed less analgesia at 10 days postpartum (8.5% vs. 13.5%; P = 0.002).

Healing and wound breakdown. Evidence was inconsistent with respect to healing and wound breakdown. The Mahomed et al. trial found no significant differences in edema, bruising, or inflammation between the groups on day 2. The trial by Kettle and colleagues reported less morbidity in the continuous suture group on day 2 for uncomfortable stitches (OR = 0.78; 95% CI, 0.64–0.96) and tight stitches (OR = 0.40; 95% CI, 0.22–0.74), but they did not find a difference in wound gaping. They reported significant differences on day 10 for wound gaping (OR = 0.46; 95% CI, 0.29–0.74), uncomfortable stitches (OR = 0.58; 95% CI, 0.46–0.74), and tight stitches (OR = 0.43; 95% CI, 0.27–0.69). The need to remove sutures was significantly lower in the continuous method group on day 10 (OR = 0.17; 95% CI, 0.10–0.28). The Mahomed trial measured this variable only at 3 months; they reported a significant difference in favor of continuous subcutaneous closure (26% vs. 37%; P < 0.001).

Incontinence and pelvic floor function. Neither trial investigated the difference in suturing methods regarding outcomes related to incontinence and pelvic floor function.

Sexual function. Neither trial found significant differences between the groups regarding dyspareunia at 3 months postpartum. The trial that measured dyspareunia at 1 year also found equivalent outcomes across the groups.

Other outcomes. Kettle et al. collected information from the women about their overall satisfaction with the repair.⁴⁸ Significantly more women in the continuous method group reported being satisfied with the repair at both 3 months and 12 months postpartum (OR = 1.64; 95% CI, 1.28–2.11 and OR = 1.68; 95% CI, 1.27–2.21, respectively).

Results for Materials for Repair

Absorbable Suture vs. Tissue Adhesive. Two trials examined use of tissue adhesive in the repair of episiotomy as compared to polyglycolic-acid⁶⁴or chromic-catgut sutures⁴⁷ (Evidence Table 5, Appendix C) Both trials were poor quality because randomization methods were either broken⁶⁴ or inadequate (odd and even registration numbers).⁴⁷ As such, conclusions about perineal morbidity related to the use of tissue adhesive are speculative at best.

Perineal pain and analgesia use. Both trials report less pain during several activities and at rest in women whose episiotomies were repaired with tissue adhesive. Bowen and colleagues⁶⁴ used a 10-point visual analog scale (VAS) and Adoni and colleagues⁴⁷ used a Likert pain scale of 1 (minimum) to 5 (maximum). Because of the differences in the pain scales, the results are not directly comparable, but the overall differences do contribute to the consistent evidence in these two trials that adhesive may lead to less perineal pain in the immediate postpartum. Both trials reported less pain while walking (1.6 vs. 2.6, P < 0.001⁴⁷ and 2.7 vs. 4.0, P = 0.0015)⁶⁴ on day 2. One trial⁴⁷ reported significantly less pain on day 3 (2.0 vs. 2.9, P = 0.029). Both trials reported less pain during micturition: in the Adoni and Anteby trial⁴⁷ on days 1 and 3 (4.5 vs. 6.3, P = 0.025 and 3.0 vs. 4.0, P = 0.025, respectively) and, in the Bowen and Selinger trial,⁶⁴ on day 2 (1.0 vs. 1.7, P < 0.031). Bowen and Selinger⁶⁴ reported less pain in the adhesive group on day 2 (1.95 vs. 3.3, P < 0.001), both while sitting (1.75 vs. 3.6, P < 0.0001) and while lying down (1.0 vs. 2.35, P < 0.001). Adoni and colleagues reported less pain in the adhesive group during defecation on days 3 and 4 (2.2 vs. 4.3, P = 0.003 and 2.1 vs. 3.7, P = 0.015, respectively). Nonsignificant differences were reported on other days. Neither trial compared need for analgesia between the two repair groups.

Healing and wound breakdown. Neither trial investigated differences in these outcomes by type of repair materials. One trial⁶⁴ did report that they identified no cases of wound infection or dehiscence.

Sexual function. Only one trial reported on sexual functioning postpartum.⁶⁴ The group repaired with adhesive, in this case Enbucrilate^® tissue adhesive, had a 35 percent reduction in the onset of pain-free sexual intercourse (P = 0.0009). Neither trial reported any other outcomes related to sexual functioning.

Absorbable Sutures: Standard vs. Rapidly Absorbed. Two studies compared standard absorbable to rapidly absorbed sutures (Evidence Table 6, Appendix C). McElhinney and colleagues, in a poor-quality trial,⁶² compared standard absorbable sutures with rapidly absorbable suture material (Vicryl® polyglycolic acid). Kettle and colleagues, in a trial of good quality,⁴⁸ also addressed the continuous versus interrupted methods of suturing in its 2×2 factorial design while comparing standard polyglactin 910 with its rapidly absorbed counterpart.

Perineal pain and analgesia use. The poor-quality trial, although it used good measurements of pain (VAS and Likert scale), found no significant differences in perineal pain between the two groups at 24 hours and 3 days. Analgesic use before discharge also did not differ by group. The good-quality Kettle trial reported mixed results for perineal pain during specific activities and the need for analgesia at 10 days postpartum. Women in the rapidly absorbed suture groups reported less pain while walking, sitting, passing urine, and defecating. Only the differences in pain with walking were statistically significant (OR = 0.74; 95% CI, 0.56–0.97; P = 0.004). Women in the rapidly absorbed group also reported less need for analgesia (8% versus 14%, P = 0.0002). This trial randomized women to suture method (continuous versus interrupted), so the investigator was able to complete stratified analyses; they showed nonsignificant results by both method of suturing and degree of trauma. In other words, improvement in pain was independent of method of closure or size of defect.

Healing and wound breakdown. The poor-quality trial combined all healing outcomes, such as infection, gaping wound, or residual material requiring removal, into one group and found that at 6 weeks, 30 percent of women whose episiotomies were repaired with standard material and 1.7 percent of women whose episiotomies were repaired with the rapidly absorbed material reported problems. The good-quality trial examined different healing outcomes separately and did not find significant differences between the two groups with respect to wound gaping, uncomfortable sutures, or tight stitches at 2 or 10 days postpartum. However, groups had meaningful differences in need for removal of sutures between 10 days and 3 months. Women whose episiotomies were repaired with the rapidly absorbed material required removal less often than women who received the standard suture at 10 days, between 10 days and 3 months, and at any point before 3 months postpartum (OR = 0.38, 95% CI, 0.23–0.64; OR = 0.19, 95% CI, 0.13–0.30; and OR = 0.26, 95% CI, 0.18–0.37, respectively).

Sexual function. Both trials measured dyspareunia, although at different time points in the short- and long-term periods of followup. The good-quality trial found no statistically significant differences at 3 months or 12 months postpartum. The poor-quality trial found that women with repairs using rapidly absorbed material had significantly lower dyspareunia scores than women who received standard sutures. This finding extended to 3 months (mean scores 0.05 versus 0.27 in women who had dyspareunia, P < 0.05) but the authors noted that the scores were very low in both groups.

Untreated Catgut vs. Treated Catgut. Only one fair-quality trial⁵⁶ investigated the use of treated, glycerol-impregnated “softgut” compared to chromic catgut. A followup to the original trial occurred at 3 years⁵⁷ (Evidence Table 7, Appendix C).

Perineal pain and analgesia use. Women in the softgut group reported significantly greater perineal pain (P = 0.015) at 10 days postpartum. Women in the softgut group were also significantly more likely to have used a perineal salt bath to relieve pain (42% versus 34%, P = 0.03) and to use more doses of oral analgesia (P = 0.18), though this difference was not statistically significant. At 3 months postpartum, self-report of perineal pain did not differ by type of catgut used.

Healing and wound breakdown. Sutures were removed more often in the chromic-catgut group both by 10 days (2.4% versus 11.5%, P < 0.001) and 3 months (6.9% versus 16.4%, P < 0.0001). Removals were described as being for maternal discomfort. Based on assessment by a midwife at 10 days, risk of perineal breakdown and healing by secondary intention did not differ by group.

Sexual function. Sexual function was assessed at 3 months postpartum and 3 years. More women in the chromic-catgut group reported pain-free sexual intercourse than women in the softgut group (50.7% versus 38.0%, P < 0.025). This difference was significant for both transient and persistent pain. At 3 years, more women in the softgut group still reported painful intercourse (OR = 1.17; 95% CI, 1.1–2.6; P < 0.02); a majority with pain described the pain as “soreness.”

Nonabsorbable vs. Absorbable. One good-quality trial, the Southmead suture study by Mahomed and colleagues,⁵⁸ and one fair trial by Buchan and Nicholls⁵⁴compare absorbable sutures for repair of the perineal skin with nonabsorbable, silk sutures (Evidence Table 8, Appendix C). The good-quality trial randomized using a 2×3×2 factorial design and randomized women to one of three material groups for suturing of the perineal skin; polyglycolic-acid, chromic-catgut, or silk sutures. Balance was obtained between use of polyglycolic acid and chromic catgut for repair of the deeper tissue and between continuous and interrupted methods of suturing. The fair-quality trial randomized women to either silk sutures or polyglycolic-acid sutures for repair of the perineal skin and chromic catgut was used in both groups for repair of the deeper tissues. In this study, method of repair of the perineal skin has potential to confound outcomes because the silk-suture group was repaired using an interrupted method and the polyglycolic-acid suture group was repaired using a subcuticular method. The trial is included under this heading because the authors frame the primary goal of the trial as a comparison of silk suture and polyglycolic-acid suture.

Perineal pain and analgesia use. The Mahomed et al. study reported no statistically significant differences among the material groups with respect to perineal pain or use of analgesia at 2 and 10 days postpartum or at 3 months. The Buchan and Nicholls trial did report differences between the groups, but the results were inconsistent by the day 6 postpartum. The investigators used the mean number of analgesic tablets used by the women to make inferences about the level of perineal pain experienced by the women. Women with repairs made with silk sutures used more analgesia than women whose episiotomies were repaired with polyglycolic-acid suture, but the results were only significant for days 3 through 5 (P < 0.001).

Healing and wound breakdown. The Mahomed et al. study reported no significant differences among the groups with regard to bruising, edema, or healing, outcomes that were clinically assessed at 2 days postpartum. In the long-term postpartum period, the silk suture group needed the absorbable suture materials removed from perineal tissue significantly less than the polyglycolic-acid or chromic-catgut groups (7 percent versus 39 percent versus 23 percent, P < 0.001).⁵⁸ No other results were reported and the fair-quality trial did not contribute to this outcome assessment.

Sexual function. In the good-quality trial, more women in the silk-suture group had not resumed intercourse by 3 months postpartum (15 percent versus 9 percent and 11 percent, P < 0.05). Among women who had resumed intercourse, dyspareunia risk was comparable. Conflicting evidence was reported by the fair-quality trial: at 4 months postpartum, more women in the silk-suture group reported no pain at all during intercourse (21 percent versus 11 percent, P < 0.001).

Polyglycolic Acid vs. Chromic Catgut. Eight RCTs compared polyglycolic-acid sutures with chromic-catgut sutures, both absorbable materials (Evidence Table 9, Appendix C). Information on these trials, ordered from the most recent to the oldest, appears in Table 7. Check marks indicate whether one type of sutures had better outcomes than the other; “ND” indicates no difference.

Both the Mahomed et al.⁵⁸ and the Mackrodt et al.^{61, 63} trials were of good quality; they contributed to other sections of this key question because of their factorial design. Four trials were of fair quality.^{50, 53, 59, 65} Finally, two trials were of poor quality;^{51, 52} both used methods of randomization that are not considered to be truly randomized; thus, we considered them to have a fatal flaw.

Perineal pain and analgesia use. All eight trials investigated differences in perineal pain outcomes between the two groups. All but two provided consistent evidence that polyglycolic-acid sutures have an advantage over chromic catgut with regards to perineal pain. Two of the fair-quality trials^{59, 65} found no significant differences between the groups; in one trial, the estimate of effect favored polyglycolic-acid suture for their pain measures on postpartum day 3 (OR = 0.70; 95% CI, 0.46–1.08). The same group of women who were in the polyglycolic-acid suture group, however, were more likely to have perineal pain at 6 months (OR = 1.77; 95% CI, 0.57–5.47), although the precision of that estimate is much less than that others in the study.

Two fair^{50, 53} and two poor trials^{51, 52} also reported less perineal pain and less need for analgesics in women who received polyglycolic-acid sutures in the short-term postpartum period. All four trials used Likert scales (mild, moderate, severe, or the equivalent) to measure pain; one trial⁵⁰ counted the proportion of women requiring analgesia (tablets or injections). Although these trials offer only fair- or poor-quality evidence, they do contribute consistent evidence that polyglycolic-acid sutures may be associated with less short-term perineal pain.

The best evidence comes from the two good-quality trials.^{58, 61, 63} In the short-term postpartum period, both trials reported pain outcomes at 24 to 48 hours and 10 days. At 24 to 48 hours, more women in the chromic-catgut group in the Mahomed et al. trial required analgesia (54 percent versus 48 percent, P < 0.05;⁵⁸ 47 percent versus 42 percent, P = 0.03^{61, 63}). This requirement for analgesia continued at 10 days postpartum in both trials and remained statistically significant in the Ipswich trial (10 percent versus 6 percent, P = 0.01). In the Ipswich trial,^{61, 63} women in the chromic-catgut group reported more perineal pain and greater severity of pain at 24 to 48 hours (P for trend = 0.002) and at 10 days postpartum (P for trend = 0.05). The earlier trial had not identified significant differences for perineal pain at these two time points.⁵⁸ Neither trial found significant differences between the groups regarding perineal pain at 3 months. In addition, the Ipswich trial found no differences at 1 year.

Healing and wound breakdown. Six of the eight trials examined healing and wound-breakdown outcomes between the two suture groups. Three fair trials^{50, 59, 65} and the Mahomed et al. study⁵⁸ did not report statistically significant differences between the groups with respect to removal of sutures, resuturing, wound breakdown, inflammation, edema, bruising, or infection, most of which were measured before 3 months. One of the poor trials⁵¹ reported significantly more edema (P < 0.05 at the perineotomy site on day 3 postpartum) in the chromic-catgut group. The Ipswich trial results favored polyglycolic-acid suture at various followup time points and on various measures. At 24 to 48 hours and 10 days postpartum, more women in the chromic-catgut group reported uncomfortable stitches (40 percent versus 33 percent, P = 0.003, and 26 percent versus 19 percent, P = 0.001, respectively). The midwives reported more “wound gaping” in the chromic-catgut women at 10 days postpartum (26 percent versus 16 percent, P < 0.00001) but not at 24 to 48 hours. By 10 days, neither group was more likely to have sutures removed, but fewer women in the chromic-catgut group reported ever having any sutures removed by 3 months (7 percent versus 12 percent, P = 0.002). This finding is the only outcome representing a disadvantage among those who received polyglycolic-acid sutures in this trial.

Sexual function. Three trials, two of good quality^{58, 61, 63} and one of fair quality,⁶⁵ investigated the effect of suture type on outcomes of sexual function. The fair-quality trial found no significant differences between the groups at 6 weeks, 3 months, or 6 months postpartum. The adjusted odds ratios (adjusted for parity) indicated a possible association between polyglycolic-acid sutures and less resumption of intercourse and more dyspareunia among women who had resumed intercourse. In one good-quality trial, women whose repairs were made with polyglactin 910, a polyglycolic acid-based suture, were less likely to suffer from dyspareunia at 1 year (RR = 0.59; 95% CI, 0.39–0.91; P = 0.002) and less likely to fail to resume pain-free intercourse (RR = 0.57; 95% CI, 0.38–0.87; P < 0.01).^{61, 63} In both of the good-quality trials, all other comparisons yielded nonsignificant differences between the two suture groups at 3 months and 1 year.

Results on Combined Approaches to Repair: Methods and Materials

Two additional trials investigated approaches to repair of perineal defects but did not distinguish between methods and materials⁴⁹ (Evidence Table 10, Appendix C).⁵⁵ Because of this, the trials are not directly comparable to other trials included in this review, and they do not contribute to the separate bodies of evidence for methods or materials. However, the results may be more applicable in the clinical setting where using a particular technique with a certain suture material, because of the properties of the materials themselves, may be more practical. We briefly describe these two trials and their results below.

One trial randomized women to one of two groups, both of which used the standard chromic-catgut approach to repair the deeper vaginal defect. In one group, the perineal skin was repaired with chromic catgut using an interrupted method of suturing; in the other, perineal skin was repaired with PROLENE™, a nonabsorbable suture material, using a subcuticular approach.⁴⁹ A similar set of groups was seen in the Buchan and Nichols trial (included above in the “Nonabsorbable versus Absorbable” section).⁵⁴ However, the Doyle trial, unlike the Buchan and Nicholls trial, was clear in its intent to investigate an entire approach to repair rather than a particular method or material.

The Doyle trial is of poor quality because it used a fair-quality randomization and implementation approach, had a small number of post-randomization exclusions, and had poor retention of subjects at followup, even in the short-term, immediate postpartum period.

As assessed by the midwife at 2 and 10 days, the groups did not differ with respect to perineal pain, need for analgesia, or bruising. The groups did not differ in pain or pain during sexual intercourse at 3 months.

Another trial randomly allocated women to one of three groups.⁵⁵ The first group had the episiotomy repaired using chromic catgut for the deep tissues and perineal muscles and an interrupted method using nylon for the perineal skin. The second group received polyglycolic-acid sutures for the deep tissues and perineal muscles and an interrupted method using polyglycolic-acid sutures for the perineal skin. The third group is described more ambiguously and had a repair done with polyglycolic-acid sutures for the deep tissues and perineal muscles and a subcuticular method using polyglycolic-acid sutures for the perineal skin.

This trial is of fair quality because of fair definitions and measurements of the outcomes and post-randomization exclusion of 98 women whom the authors were unable to follow because of relocation. These exclusions were nondifferential across the three randomized groups.

More women who had a repair with subcuticular polyglycolic suturing had no discomfort at 5 days than did women who had repairs with interrupted nylon or interrupted polyglycolic-acid sutures (40%, 12%, and 18%, respectively, P < 0.001). The groups that had repairs with the interrupted method did not differ. The authors reported significant differences between women who had repairs with subcuticular polyglycolic-acid sutures and the interrupted method group including pain during sitting, walking, and bowel movements at 5 days. As assessed by the midwife at 5 days, the group repaired with subcuticular polyglycolic-acid sutures experienced less edema (11 percent versus 30 percent versus 23 percent, P < 0.005). No significant differences were found with respect to infection or hematoma.

At 3 months, the polyglycolic-acid suture groups differed significantly. Women whose episiotomies were repaired with the subcuticular approach were less likely to suffer from dyspareunia, discomfort with defecation, incontinence of flatus or discomfort when sitting (P < 0.025) than women whose episiotomies were repaired with the interrupted approach.

Literature Search and Included Studies

Overview. We identified 16 publications that prospectively collected data about some aspect of continence or pelvic floor muscle function with good documentation of perineal status and episiotomy use at the time of the index birth. Outcomes of interest included physiologic measures of muscle strength, clinical urodynamic testing, or self report by interview or questionnaire. No studies directly compared type of incision and future pelvic floor function. The 16 publications include four reports from two randomized clinical trials (RCTs) of liberal versus restrictive use of episiotomy, 11 prospective studies of representative cohorts of women delivering at particular facilities or with a particular practice group (including two publications from a cohort of women who participated in an RCT of perineal massage versus none in the third trimester), and one cohort composed of all women in a region who had third-degree lacerations at the time of the index birth. The last study followed the cohort to assess risk of fecal incontinence at 3 months.

Two publications came from the same population in the United Kingdom: a primary analysis of the RCT outcomes at 3 months,²³ and a secondary analysis after 3 years of followup.³³ Two Canadian reports also present analyses of the same study population.^{29, 34} In this case, both publications report 3-month followup data: one analysis by the initial trial groups of liberal versus restrictive use of episiotomy and the other an analysis based on classification of perineal trauma at the index birth. Of the remaining 12 prospective studies, three were conducted in the United Kingdom, two in Denmark (separate populations), two in Canada (separate populations), two in Sweden (separate populations), one in Italy, one in Turkey, and one in the United States. In total, the 16 publications represent 12 unduplicated study populations from seven countries.

Study Participants. All but the study of third-degree lacerations restricted study participants to those with term, singleton gestations at the time of the index birth. Four of 16 studies restricted their study populations to primiparous women;^68–71 four studies restricted to spontaneous vaginal deliveries.^{23, 33, 71, 72} To assess what component of change in pelvic floor muscle function could be attributed to pregnancy and what component was most influenced by vaginal births, two studies included a nonpregnant comparison group and a group of women who had cesarean birth in their physiologic measures.^{68, 73} In each case the comparison groups were recruited contemporaneously to the women who had vaginal deliveries generally from the same practice.

Episiotomy Type. We did not identify any studies that compared the influence of mediolateral versus midline (also called median) episiotomy on pelvic floor function or continence. The remainder of the studies reflects the dominant practice patterns in the countries in which the studies were conducted. Mediolateral episiotomy was the rule (with very rare exceptions) in European and Turkish cohorts; midline episiotomy predominated in the United States and Canada. This implies that to some degree European and North American studies are investigating fundamentally different exposures. The anatomic location, involved tissue planes, extent of perineal body disruption, and risk for extension associated with mediolateral compared to midline episiotomy would be expected to be distinctly different.

A single randomized trial comparing the two methods has documented that the risk of extension into and/or through the rectal sphincter is more than 2.5 times more likely with midline episiotomy; sphincter involvement occurred in 24 percent of deliveries with midline episiotomy in their trial.³⁶ They also noted that local extension, not involving the sphincter, was 1.8-fold more likely with midline episiotomy. These differential outcomes are believed to represent differences in the tissue planes involved. They may also represent differences in familiarity with midline episiotomy technique since reported risk of extension is lower in prospective cohorts in countries where midline episiotomy is routine. Because no trials or prospective studies directly compare the pelvic floor muscle function across type of episiotomy, the long-term differences in continence and pelvic floor muscle outcomes that would be anticipated secondary to differences in the type is unknown. These differences must be taken into consideration when synthesizing the findings relevant to this key question.

Outcomes Measured. This question was aimed at identifying research publications that undertook long-term followup, measured in years. However, we identified only five publications from four study populations with followup of a year or longer.^{33, 68–70 74} In response, we have included the entire literature that assesses continence and pelvic floor function at any time after the arbitrary 8-week window that can be used to define the postpartum period. Shorter-term continence and pelvic floor outcomes in the days and weeks around birth are described in the section on the outcomes of routine use of episiotomy.

To summarize outcomes, we grouped measures into four categories: those that assess urinary incontinence by self-report, those that assess continence of stool and flatus by self-report, those based on physical examination findings to describe anatomy, and those measures intended to document physiologic function, such as perioneometry. Seven studies assessed urinary incontinence by self-report; timeframes for self-report included 3 months,^{23, 29, 34, 70, 71, 75, 76} 12 months,⁷⁰ 3 years,³³ 4 years,⁷⁴ and 5 years.⁶⁹ Three studies assessed continence of stool and/or flatus by self-report at 3 months,^{71, 75, 77} and one study collected self-report data at an average of 10 months postpartum. Two studies described physical examination findings related to prolapse⁷¹and anal sphincter function and anorectal anatomy.⁷⁷ Five studies used perineometry measures to document characteristics of muscle function such as maximum strength of contraction and maximum sustained contraction over 10 seconds.^{29, 34, 68, 71, 73} A single study reports findings from urodynamic testing that included observation of stress incontinence with strain and timing of interval required to stop urine flow;⁷¹ one study used weighted vaginal cones recording the heaviest weight that could be retained while standing or walking antepartum and 2 months postpartum. The highest-quality studies combined types of measures and reported data in standardized fashion that concurs with definitions of the International Continence Society.

Quality. Quality assessment is described in detail in Chapter 2 (Methods); key components of quality assessment and an overall quality score are provided in Evidence Table 11 (Appendix C). Four publications for this key question derive from the conduct of two RCTs that had good-quality ratings for assessing outcomes of liberal versus restrictive use of episiotomy. For this question, we also rated the breadth of measures used to characterize outcomes, the clarity of specification of the outcome measures (including documentation of how participants were asked self-reported measures), use of measures with documented validity and reliability, and loss to followup. For prospective cohort studies, we assessed these features and the representativeness of the participants to reflect a base population of women having births, as well as use of adjusted models to control for potential confounding factors.

Table 8 summarizes the methods and findings of the individual studies identified. Publications are listed in order from older to more recent reports. We then separately consider the findings from randomized trials of liberal versus restrictive use of episiotomy and prospective studies that employ episiotomy or perineal trauma categories as the primary exposure of interest. The findings are further grouped within study type by urinary incontinence outcomes, rectal continence outcomes, anatomic findings, and pelvic muscle function measures obtained using physiologic measurements.

Results

Randomized Clinical Trials. Both randomized clinical trials, Sleep and colleagues in the United Kingdom²³ and Klein and colleagues in Canada,²⁹ conducted trials that required providers to alter their use of episiotomy. These trials randomized women to receive “liberal use” versus “restricted use” of episiotomy, with the latter category intended to restrict use to circumstances such as fetal distress or maternal exhaustion with an “unyielding perineum.” Both trials enrolled singleton, vertex presentation pregnancies at term and randomized in the delivery suite close to the time of birth.

The United Kingdom trial had a 10.2 percent use of episiotomy in the restrictive group (2.6 percent for maternal indications; 6.6 percent for fetal distress), compared to 51.4 percent use of episiotomy in the liberal group. The Canadian trial had greater difficulty modifying provider behavior as background rates of episiotomy exceeded 80 percent. Restrictive use resulted in 57.2 percent of the women having an episiotomy compared to 81.4 percent in the liberal-use arm. Each of these research groups published an analysis as randomized for 3-month postpartum data. The sole violation of intention to treat was Klein's elimination of five women with cesarean section from analysis of pelvic floor outcomes. Both trials achieved good balance of baseline characteristics through randomization. Both trials assessed urinary incontinence outcomes; neither assessed continence of stool or flatus.

The Canadian trial also assessed self-reported sensation of perineal “bulging” and conducted perineometry. The Canadian team has published two analyses: an analysis as randomized and an analysis by perineal trauma sustained. The randomized analyses produced no meaningful differences in self-reported urinary incontinence, subjective sensation of perineal bulging, or perineometry readings, including when baseline antepartum readings and parity were incorporated. Likewise, the analysis by perineal status (intact, episiotomy, spontaneous tear, third- and fourth-degree tear) revealed no differences in self-reported incontinence or in perineometry scores. These analyses were stratified by parity and suggest some effect modification; however, the authors did not provide adjusted models or note any statistically significant results.

The research team from the United Kingdom has published two analyses as randomized: a 3-month and 3-year followup, both conducted by mailed questionnaire followup.^{23, 33} The 3-month followup found no difference by group in risk of involuntary loss of urine (19 percent in both arms) or in need to wear a pad because of urine loss (6 percent in both arms). Their 3-year followup was more detailed and included involuntary loss of urine; use of a pad; loss of urine with coughing, sneezing, laughing; and loss with urgent need to void. No aspect of these symptoms or their severity varied by restrictive versus liberal episiotomy group. For the 3-year followup, this lack of difference across groups persisted when taking into account subsequent obstetric history. The 3-year followup was also marred by loss, although the authors were able to use 3-month data to demonstrate little evidence of response bias. Adjustment using multivariable models is alluded to but numeric data are not provided.

Neither trial collected data about continence of flatus or stool, descriptive data from physical examination, or urodynamic studies. Both research teams concluded that they did not observe any benefits associated with episiotomy. Klein and colleagues, based on perineometry measures, also concluded that episiotomy fails to prevent pelvic floor relaxation.

Prospective Studies. The Italian study by Sartore and colleagues provided the most global assessment of continence and pelvic floor function; they addressed each of our four categories of outcomes.⁷¹ They enrolled 519 primiparous women who had singleton, spontaneous vaginal births in lithotomy position. Women with pre-existing incontinence were excluded. Measures of outcomes at 3 months included in-person interviews, physical examination, perineometry, a test to provoke stress urinary incontinence, and a urine-stream-interruption test. The study team clearly described methods, used a standard scheme for classifying prolapse, and collected data about urinary and anal incontinence. Overall measures and implementation were good. For the entire panel of outcomes (stress urinary incontinence, anal incontinence, anterior prolapse, posterior prolapse, vaginal manometry, and urine-stream interruption), there was only one statistically significant difference in perineometry findings. Women who did not have an episiotomy (all mediolateral) had higher contraction strength on perineometry (13.8 compared to 12.2; P < 0.001); moreover, the proportion of women with abnormal manometry was higher among women with episiotomy (40.6 percent compared to 27.7 percent without episiotomy). The adjusted relative risk for abnormal manometry was 1.8 (95% CI, 1.2–2.6). The study team concluded that episiotomy is associated with lower pelvic floor muscle strength than spontaneous tears. All self-reported symptoms of urinary and anal incontinence and degree of prolapse on physical examination were equivalent across groups so the clinical significance of this finding is unclear. Overall interpretation must be that episiotomy does not protect against incontinence, prolapse, or decrements in pelvic floor muscle function by 3 months postpartum.

Studies focused on self-reported urinary continence. Excluding the clinical trial populations and the study by Sartore et al. described above, five studies (in four study populations) evaluated self-report of urinary continence.^{69, 70, 72, 74, 76} Two used a telephone interview,^{70, 72} the other three mailed questionnaires.

Karacam and Eroglu provided the least-detailed information:⁷² no details about how stress incontinence was queried or defined for data analysis and no report of adjusting for factors that might influence outcomes by using stratified analyses or multivariable models. They reported, from bivariate data at 3 months (N = 100), that 24 percent of women with episiotomy and 30 percent of women without episiotomy had stress incontinence.

Eason and colleagues asked about occurrence and frequency of “involuntary loss of urine when coughing, sneezing, laughing, or running” in a cohort of 949 women also at 3 months.⁷⁶ For analysis, they reported that any stress incontinence occurred in 29 percent of those with episiotomy and in 35 percent of those without. Multivariable models for stress urinary incontinence comparing episiotomy to no episiotomy yielded an odds ratio of 0.68 (95% CI, 0.47–1.01).

Viktrup and colleagues reported using a questionnaire to obtain all the facets of the International Continence Society definitions of incontinence from 305 women.⁷⁰ They reported no differences but do not provide numeric data. In summary, they stated that although women with episiotomy had more incontinence postpartum, differences had resolved by 3 months and remained equivalent at 1 year. Their followup survey at 5 years revealed in multivariable models that episiotomy in a first birth was significantly associated with stress incontinence.⁶⁹ An adjusted point estimate is not provided.

Rockner's followup of a cohort of 185 women with either episiotomy or spontaneous tear at 4 years after the index pregnancy asked women about symptoms before, during, and after all pregnancies; information about the index pregnancy focused on frequency and severity of urinary incontinence; need for pad; and loss of urine provoked by cough, laugh, or sneeze.⁷⁴ Symptom profiles were very similar across groups; for instance, 34 percent of women who had an episiotomy and 30 percent without reported incontinence provoked by cough, laugh, or sneeze. No stratified or adjusted models are provided. The author concluded that episiotomy and spontaneous-tear groups had the same frequency of incontinence symptoms. Overall, each research team investigating self-reported urinary incontinence concluded that no evidence supported the view that episiotomy prevents pelvic floor defects.

Studies focused on self-reported incontinence of stool or flatus. Three cohort studies asked women about rectal incontinence symptoms.^{75, 77, 79} One study also conducted physical examinations.⁷⁷

The earliest of these studies was conducted on a cohort constructed of 81 women who had third-or fourth-degree lacerations.⁷⁷ The prevalence of episiotomy among women without third- or fourth-degree lacerations was known as well as their episiotomy history in the index pregnancy. They were followed up at 3 months when their symptoms were evaluated and they received a physical examination. All women who had fecal incontinence and abnormal rectal examination at three months had had an episiotomy. Of those with an abnormal exam and no incontinence 60 percent had had an episiotomy, meaning relative risk of abnormal exam was 50 percent greater among those with history of episiotomy. These authors focused on the high prevalence of anorectal dysfunction at 3 months with episiotomy as a key risk factor. None of the research teams that focused on incontinence of flatus or stool found episiotomy to be significantly associated with reduced risk.

MacArthur and colleagues sent questionnaires at 6 to 7 months and then followed up all women who had a variety of symptoms at 10-months with an in-person interview. ⁷⁹ Their questionnaire and interview classified several types of fecal incontinence and staining (not including simple flatus); any one or more of the symptoms was considered evidence of incontinence. At 10 months, episiotomy was not an independent predictor of fecal incontinence in multivariable models. Because episiotomy was not a key focus of their analysis, they do not provide a point estimate.

Eason and colleagues inquired about involuntary loss of stool or flatus and the frequency at 3 months.⁷⁵ Women with episiotomy reported higher prevalence of loss of stool (5.4 percent) and loss of flatus (30.2 percent) than did women without episiotomy (2.5 percent and 24.4 percent, respectively). Adjusted models revealed that episiotomy without extension was associated with a relative risk of 1.3 (95% CI, 0.9–1.8), and third- to fourth-degree lacerations (virtually all after episiotomy) were associated with 2.1-fold increased risk of anal incontinence (95% CI, 1.4–3.1). The study team concluded that anal incontinence was associated with severe lacerations that are most likely to result from episiotomy.

Studies focused on physiologic measures of pelvic floor function. In 1985, Gordon and Logue published the first use of perineometry to evaluate prospectively a group of 70 women.⁶⁸ They included a nonpregnant and a cesarean comparison group to take into account changes associated with pregnancy and labor, respectively. They did not compare women with their own measures in pregnancy. No difference in maximum contraction strength was seen across groups. However, the researchers did note that postnatal exercise level was highly associated with perinatal muscle strength.

Fleming and colleagues refined the Gordon and Logue study design.⁷³ They conducted perineometry antepartum, at 6 weeks, and again at 6 months among 102 women with singleton spontaneous vaginal births. Detailed measurement protocols are provided and their analysis focused on mean difference between antepartum scores and 6-month scores. No differences in perineal muscle strength or endurance were identified between laceration and episiotomy groups.

In another approach to measuring muscle strength, Rockner and colleagues conducted studies with weighted vaginal cones at 36 weeks gestation and again at 2 months postpartum. They calculated decrements in weight that could be retained while standing or walking: women with episiotomy had the greatest decrement in function (30 gm decrease in maximum weight held), compared to 19.2 gram decrease with intact perineum, and 18.9 gm decrease with spontaneous tears (P < 0.001).⁷⁸

Overall, none of these research teams concluded that episiotomy had advantages, and one identified a decrease in functional muscle strength. These intermediate findings concur with the self-report and clinical examination findings of other studies that detected no evidence of benefit from episiotomy with respect to preserving continence or pelvic floor muscle function.

Literature Search and Included Studies

Overview of the Evidence. Nine publications were identified that prospectively collected outcome data about sexual function among women who did or did not have a routine episiotomy. Evidence Table 12 in Appendix C provides details. One study compared incision type and assessed sexual function.³⁶ These 10 publications include three randomized trials of restrictive versus liberal use of episiotomy;^{23, 29, 33} one trial of median versus midline episiotomy;³⁶ and five prospective cohort studies.^{34, 71, 72, 80, 81} One study (the only study conducted in the United States), described by the authors as “retrospective,” included a single followup time point (6 months) for which the data collection about sexual function was prospective.³⁸ Two publications reflect a primary analysis from a randomized clinical trial (RCT) with 3 months of followup²³ and a secondary analysis after 3 years of followup³³ in the same study population from the United Kingdom. Two publications by a Canadian research team are also analyses of the same study population.^{29, 34} In this case, both publications report 3-month followup data: one analysis focused on randomization to liberal versus restrictive episiotomy groups, and the other took the perspective of exposure groups classified by perineal trauma at the time of delivery. Two of the prospective studies were conducted in Sweden (separate study populations), one in Italy, and one in Turkey. Thus, in total, this literature represents seven distinctive study populations from six countries.

Study Participants. In current practice in the United States, women who are giving birth for the first time are most likely to have a routine episiotomy. Several studies that evaluated sexual function restricted the study population to primiparous women. This approach assures that the influence of episiotomy, spontaneous laceration, or intact perineum reflects only the potential influences of the index birth, rather than both the index birth and any prior history of perineal trauma among women who have had prior births. Those studies that did not restrict their study of sexual function to primiparous patients adjusted for prior episiotomy in data analysis as a method to account for the influence of prior birth experiences. All studies restricted participation to singleton births; and some specifically included only women who had a spontaneous vaginal birth.

Episiotomy Type. Only one study directly compared mediolateral to median (midline) episiotomy.³⁶ The remainder of the studies reflects the dominant practice patterns of the countries in which the studies were conducted. Mediolateral episiotomy is routine in the countries represented, with the exception of the United States and Canada, where median episiotomy is routine. Overall, when episiotomy was performed in the U.S. and Canadian studies, it was a median incision;^{29, 34, 38} this phenomenon stands in contrast to 98.9 percent mediolateral episiotomies in the European and Turkish studies.^{72, 33, 71, 80, 80, 81}

This factor introduces a fundamental difference in the “exposure” across studies. The anatomic location, involved tissues plains, extent of perineal disruption, and risk for extension associated with mediolateral as compared to median episiotomy are distinctly different. Once healed, the scar from each type of episiotomy and from spontaneous lacerations will be subject to different amounts and types of contact, pressure, and stretch depending on position of partners during sexual intercourse. Thus, the literature reflects two distinct types of procedures, the effects of which need to be addressed separately.

Outcome Measures. Of the 10 studies included for this key question, eight were not designed to address sexual function as the primary outcome. Only the Signorello et al. and Karacam and Eroglu studies reported that a primary objective of the study was to assess the relationship between perineal trauma (spontaneous versus episiotomy) and postpartum sexual function.^{38, 72}

The most consistently reported outcome was “dyspareunia.” In three of these 10 publications, the researchers provide no detail to document how they phrased a question or questions about pain with intercourse or how they recorded participant responses; no reports distinguished between pain on insertion, deep dyspareunia with thrusting, or residual pain (deep or perineal) after intercourse. Four studies used a written questionnaire to collect information about sexual function.^{23, 33, 38, 80} One study conducted telephone interviews,⁷² and four conducted in-person interviews.^{29, 34, 71, 81} In-person interview methods for assessing sexual function outcome tended to be more detailed than those obtained from written questionnaires. However, none of the publications distinguish between pain on insertion, deep dyspareunia with thrusting, or residual pain (deep or perineal) after intercourse.

Across all 10 studies, investigators used three approaches for summarizing when women experienced dyspareunia. The most common was to inquire about any dyspareunia since resuming intercourse. Other authors inquired about dyspareunia with episodes of intercourse near the time of the followup. To differentiate this approach from measures of any experience of dyspareunia, we have called inquiry about recent status “current dyspareunia” in this report. Less often, authors reported about pain at the time of the first episode of intercourse after the index birth.

In a related measure, four research teams also asked women to recall when they resumed having intercourse. This question allows the investigators to report both continuous and categorical data about the proportion of women who had resumed intercourse by particular points in time, for example, by 2 months postpartum.^{23, 29, 34, 80} Few authors clearly explained if the prevalence of dyspareunia reported is appropriately calculated as a proportion (number of women with pain with intercourse divided by number of women who have resumed intercourse). The most common timeframe for assessment of outcomes was 3 months. One group assessed dyspareunia at postpartum exams between 2 and 3 months;⁸¹ one used a mailed questionnaire at 6 months;³⁸ and the longest followup was conducted by questionnaire mailed at 3 years.³³

The two publications by Klein and colleagues had the most elaborate approach to collecting several types of information. These authors reported greater detail about how participant responses were collected and analyzed. In interviews at 3 months postpartum, they asked women when they resumed intercourse and assessed recalled pain at the first postpartum episode of intercourse using the McGill Pain Scale. They inquired about sexual satisfaction using an unspecified number of items measured on a 4-point scale and reported a summary measure of “sexual satisfaction” in their tables.^{29, 34} Two other groups classified degree of pain with intercourse using an approach that assigned levels: none, mild, moderate, and severe.^{71, 34}

Quality. None of the identified studies was designed exclusively to examine sexual function. We classified primary and secondary outcomes based on objectives provided in the introduction of the publication or used stated research questions to classify primary and secondary objectives.

None of the 10 studies met criteria that we consider necessary to be a good study of sexual function after episiotomy. Our criteria included (1) documentation of a representative sample of women who had spontaneous vaginal births, (2) use of outcomes that provide a well-rounded picture of sexual function, (3) clear specification of outcome measurement approach (including specification of items asked of participants on surveys or in interviews), (4) use of measures with documented validity and reliability, (5) use of adjusted models in prospective data to control for potential confounding factors, (6) minimal to modest loss to followup, and (7) use of intention-to-treat analysis in randomized clinical trials.

Results

Table 9 summarizes the methods and findings of the individual studies identified. Publications are listed in order from older to more recent reports. We also separately consider the findings of controlled trials of liberal use versus restricted use of episiotomy and other prospective cohort studies of episiotomy that include sexual function outcomes (Table 10). The single study that compared type of episiotomy incision and included assessment of sexual function is reviewed on pages 30 and 31; no differences in pain with initiation of intercourse or with satisfaction with intercourse were noted by episiotomy type.

Randomized Controlled Trials. Two publications present results from RCTs of restrictive compared to liberal use of episiotomy; the investigators used an intention-to-treat analysis. These trials provide evidence about the long-term effects of a particular type of policy about episiotomy use on the sexual outcomes of populations of women. The earlier of the two trials was conducted in the United Kingdom in 1982.²³ Perineal outcomes in the West Berkshire Perineal Management Trial differed clinically and statistically by group. Among women in the liberal-use group, 51.4 percent had had an episiotomy (all mediolateral), 6.0 percent had an episiotomy with extension to third- or fourth-degree laceration, 24.5 percent had a spontaneous perineal tear only, and 24.3 percent had no perineal trauma. In the restrictive-use group, 10.2 percent had an episiotomy, 1.2 percent had an episiotomy with extension, 55.8 percent had a spontaneous perineal tear only, and 33.9 percent had no trauma. By 1 month after delivery, 37 percent of the restrictive group and 27 percent in the liberal group had resumed sexual intercourse (P < 0.01). The proportion of women with resumption of intercourse by 3 months, current dyspareunia at 3 months, or any dyspareunia within the 3 months of followup did not differ significantly by group.²³ By the third year of followup, the likelihood of “ever suffering painful intercourse” remained comparable across groups.³³

The trial conducted by Klein and colleagues in Canada also found less episiotomy use in the restrictive group with higher rates of spontaneous lacerations.²⁹ Among women in the liberal-use group, 67.2 percent had an episiotomy (midline), 14.2 percent had an episiotomy with extension to third- or fourth-degree laceration or sulcal tear high in the vaginal vault, 12 percent had a spontaneous perineal tear only, and 6.6 percent had no perineal trauma. In the restrictive-use group, 42 percent had an episiotomy, 15 percent had an episiotomy with extension, 35 percent had a spontaneous perineal tear only, and 7.5 percent had no perineal trauma. Women in the restrictive group resumed intercourse an average of 1 week earlier that those in the liberal group; however, all other measures of sexual function were equivalent by 3 months.²⁹

These trials were designed primarily to assess rates of episiotomy and perineal trauma under different strategies to guide use of episiotomy. Restrictive use, as addressed in KQ 1, was hypothesized to result in less severe trauma among women with lacerations and in a higher proportion of women without perineal lacerations. If women experienced less perineal trauma, this improvement would be expected to be associated with less pain with future intercourse. Therefore, the Canadian trial team also undertook a separate analysis of the relationship between the degree of perineal trauma and sexual function. Using data from the 3-month interviews, they regrouped participants by perineal status that was systematically documented at the time of the index birth, creating a prospective cohort. In this cohort analysis, women with an intact perineum were most likely to have resumed intercourse by 6 weeks (76.5 percent), followed by those with spontaneous tears (62.5 percent), episiotomy alone (61.7 percent) and third- and fourth-degree lacerations (55.4 percent). Women with episiotomy had the slowest return to intercourse. Pain with the first intercourse followed a similar pattern.³⁴

Prospective Cohorts. Signorello and colleagues were the sole research team from the United States to assess sexual function.³⁸ They documented trauma at the time of childbirth by chart review and followed up women at 6 months. They reported that the degree of trauma, rather than whether it resulted from episiotomy or spontaneous tear, was the primary determinant of pain with intercourse at 6 months. In prospective 6-month data, the risk of pain with intercourse was higher among those with second-degree trauma compared to no trauma (RR 1.3, 95% CI, 0.8–2.2), and highest with third- and fourth-degree trauma (RR 1.5; 95% CI, 0.7–3.5), although not statistically significant.

These cohort studies do not find large or statistically significant differences in sexual function. Only one study identified lasting differences in dyspareunia at 3 months. Sartore and colleagues reported that women with episiotomy were more than twice as likely to have pain than those without episiotomy.⁷¹ An aggregate estimate for current dyspareunia at 3 months can be estimated from three of the cohort studies using 818 women with episiotomy and 938 women without episiotomy.^{71, 80, 81} We used meta-analysis techniques to calculate an aggregate risk ratio for the combined population of the prospective cohorts. In these studies, women with episiotomy were 54 percent more likely to have pain with intercourse 3 months after delivery (RR: 1.54, 95% CI: (I: 1.19, 2.00), with an absolute increase in risk of dyspareunia of 5 percent among women who had episiotomy: 14.3 percent versus 9.3 percent. Similar estimates for the two studies that assessed any dyspareunia during the 3 months after childbirth reveal no difference in the overall probability of having had painful intercourse. Among 50 women with episiotomy, 65 percent have had pain with intercourse and 50 women without episiotomy, 54 percent had had pain with intercourse but this was not statistically significant.⁷²

Key Question 1: Episiotomy and Maternal Postpartum Outcomes

This literature, spanning two decades from the mid-1980s to the present, has high internal consistency with respect to the postpartum effects of routine (or liberal) versus restrictive strategies for episiotomy use.^{22, 23, 28–32} We found few if any meaningful discrepancies in findings, and overall we regarded this body of evidence as Grade II.

Across studies, women in the restrictive-use groups had less severe posterior perineal trauma, more frequent but not severe anterior vaginal trauma, less overall need for suturing, and higher probability of having an intact perineum when compared to routine- or liberal-use policies. These differences in perineal trauma were associated with less pain in the short term among those in restrictive-use groups, with fairly prompt resolution of pain regardless of trial arm. Women in restrictive use arms had no greater or lesser risk of wound healing complications and were more likely to resume intercourse earlier. Overall loss to followup was pronounced during the weeks after birth; remarkably little is known about recovery trajectory or complications thereafter.

Although these trials are of fair to poor quality overall, we base our conclusion that this evidence does not support routine use of episiotomy on the notable consistency of findings. Routine episiotomy does not achieve any of the short-term goals it has been hypothesized to achieve. Indeed, routine use is harmful to the extent that it creates a surgical incision of greater extent than many women might have experienced had episiotomy not been performed.

Key Question 2: Episiotomy Incision Type and Maternal Morbidity

On the question of midline vs. mediolateral episiotomy, only a single study found that women who had a midline episiotomy had a significantly greater rate of anal sphincter injuries than women in the mediolateral episiotomy group.³⁶ Treatment groups did not report differences in pain or satisfaction with intercourse at 3 months after the intervention.

Because of considerable methodological flaws in this trial (poor internal validity), any conclusions must be drawn cautiously, and we rate this “body” of evidence as Grade III. However, because differences in sphincter injury rates are clinically important, we consider the finding of increased risk of severe injury with midline episiotomy compared to mediolateral to be relevant observational evidence.

Key Question 3: Repair of Perineal Defect and Maternal Morbidity

Methods. Because of the heterogeneity of methods used for repair of episiotomies and perineal lacerations, overall conclusions are applicable only to the repair method under study. Generally, we rate the strength of evidence for these issues as Grade III except in one instance, mentioned below.

Two trials^{60, 63, 66} studied a two-layer approach, in which the perineal skin is left unsutured, against a three-layer approach, in which the skin is sutured closed. Two trials^{48, 58} investigated the differences between a continuous (or subcuticular) method and an interrupted (or transcutaneous) method. Specific limitations of the trials and conclusions can be found below, grouped by the particular method comparison under study. Overall limitations in this body of evidence and suggestions for future research are presented later in this chapter.

Two-layer vs. three-layer repair. Both trials provided consistent evidence that favored the two-layer approach, although statistically significant differences between the two approaches were not always found. The Ipswich Childbirth study had an overall good quality with adequate definitions and measurement of outcomes, reported masking of the outcome assessors and patients, and less than 15 percent loss to followup at 1 year.^{60, 63} Definitions and measurement of the outcomes were only fair in the other trial,⁶⁶ making it harder to compare to other trials. The latter trial also reported outcomes only on women who had completed all followup assessments; it did not report on possible differences between completers and the women who missed one or more followup assessments.

Because the two-layer approach involves less suturing, it also means less inflammation and bruising; this in turn could result in less pain and perineal morbidity. Another explanation for the differences may lie with suture type. In the first trial, which produced fewer significant differences between the approaches, the investigators balanced polyglycolic acid and chromic catgut sutures.^{60, 63}In the other trial, a portion of the significant differences between the groups may be explained by an imbalance in suture type used. Chromic catgut was predominantly used and is hypothesized to be associated with increased pain, edema, and inflammation.⁶⁶

Even with the limitations discussed above, the pool of evidence from both trials suggests that less overall perineal morbidity is associated with the two-layer repair approach, which leaves the perineal skin unsutured, than with the three-layer approach. The reduction in pain, need for analgesia, wound healing problems, and sexual morbidity, as well as a decrease in the time and cost required for initial suturing of the perineal skin, removal, and possible resuturing, may make the two-layer approach more beneficial than the traditional three-layer approach.

Continuous vs. interrupted sutures. Although the evidence is unclear, it suggests that a continuous method of repair, though it may be technically more difficult, may be superior to the interrupted method. Two good-quality trials produced inconsistent evidence that the continuous method of repair has less perineal morbidity and more patient satisfaction associated with it than the interrupted method of repair.^{48, 58}

Both trials, through a factorial design of randomization, also randomized women to different suture material groups. Both trials achieved valance with respect to suture type; authors represented results for methods of repair regardless of suture type. Both trials defined and measured outcomes well and achieved good followup. In both trials, the authors describe greater familiarity with the interrupted method of repair, which is said to be technically easier to perform than the continuous method. One clinical group (for the trials conducted in Southmead, United Kingdom) even suggests that their inconsistencies with other trials might have be attributable to the lack of practice with the method and subsequent unpopularity with the operators that performed the repair.⁵⁸ Whether such differences in outcome arise for clinicians and women outside the United Kingdom, where methods of repair and training of those performing the repair could be different compared to other countries, remains to be seen.

Materials. Because this review includes trials dating back to 1974, the materials used differ over time. Two trials^{47, 64} compared absorbable sutures with tissue adhesive; two trials^{48, 62} compared absorbable sutures with their rapidly absorbed versions; one trial⁵⁶ compared untreated with treated catgut; two trials^{54, 58} compared nonabsorbable sutures with absorbable sutures; and eight trials^{50–53, 58, 59, 61, 63, 65} compared polyglycolic acid with chromic catgut. Because of this heterogeneity, specific limitations of the trials and conclusions can be found below, grouped by the particular material comparison under study. Overall limitations in this body of evidence and suggestions for future research can be found later in this chapter.

Absorbable vs. tissue adhesive. Both trials^{47, 64} were of poor quality because the method of randomization was inadequate or broken. However, even though sample size was small (n < 65 in both trials), both groups did define and measure perineal pain well and achieved good followup. These trials contribute possible evidence that repair with tissue adhesive may decrease perineal pain experienced in different situations in the immediate postpartum. This conclusion must be weighed in light of the inadequate randomization of these studies. Our review suggests that this question merits further study in a well-randomized trial.

Absorbable sutures: standard vs. rapidly absorbed. The mixed results from the good trial⁴⁸ and lack of significant differences between groups in the poor trial⁶² suggest that evidence is insufficient about any difference in perineal pain between standard and rapidly absorbed sutures. We saw stronger evidence that women who had rapidly absorbed sutures required less removal of the material, presumably because it had been absorbed more quickly in the postpartum period. We had difficulty assessing the effect of type of absorbable suture on other healing outcomes because the poor trial grouped all outcomes together. Although the two trials evaluated sexual functioning at different time points, evidence suggests that rapidly absorbed sutures may decrease the amount of dyspareunia and the severity thereof in the puerperium. One item to note in the good-quality trial was the masking of the suture material. The trial acquired undyed sutures direct from the manufacturer, thereby achieving a very high level of internal validity and decreasing the amount of bias in assessment of the outcomes.

Untreated catgut vs. treated catgut. Only one trial addressed treated and untreated catgut.⁵⁶ This trial achieved fair randomization and was able to blind the assessors and the patients. Loss to followup at 10 days and 3 months was minimal, but at 3 years, loss to followup was only fair (70 percent). A small amount of crossover to the other suture material group occurred, but the investigations did perform an intent-to-treat analysis.

This trial produced no evidence that “softgut” (i.e., treated catgut) is superior to untreated catgut with regard to perineal morbidity. In fact, the trial may indicate that softgut may be associated with higher morbidity, as there appeared to be more perineal pain in the immediate postpartum period and more painful sexual intercourse in the longer-term period. Though untreated catgut sutures needed to be removed more often, the authors attributed the difference to the tendency for the sutures to dry out. However, they speculated that such drying out could not completely explain the differences in perineal pain.

The women were repaired using different techniques, but the randomized groups were balanced in that respect. Investigators used the interrupted method approximately 60 percent of the time in both groups. Stratifying by technique of repair showed more marked dyspareunia and need for suture removal in women who were repaired using interrupted sutures, a finding that is consistent with other trials investigating method of repair

Nonabsorbable vs. absorbable. Because of the study design of the fair-quality trial⁵⁴ and lack of control for possible confounding by method of repair, we cannot draw conclusions about the role of silk sutures in perineal morbidity from this trial. The authors present data by suture material and then do not mention differences in the methods until the conclusions section of their article. They concluded that the subcuticular method lent itself to short-term advantages, but they did not present the data to support their conclusion. Thus, although this trial may contribute to a body of evidence that looks at combinations of materials and methods, it does not contribute to the overall understanding of the role of suture materials in perineal morbidity, separate from methods of repair. The Mahomed et al. trial⁵⁸ found no differences between the two groups in the short-term postpartum period, but did find differences at 3 months, indicating a possible delayed effect of the suture material.

Polyglycolic acid vs. chromic catgut. In 2004, the Cochrane Library published a systematic review and meta-analysis of information on polyglycolic acid versus catgut suture material for repair of perineal trauma.⁸² In it, they report that polyglycolic acid sutures were associated with less pain in the short-term postpartum period (Odds ratio [OR] = 0.62; 95% confidence interval [CI], 0.54–0.71) and with less need for analgesia (OR = 0.63; 95% CI, 0.52–0.77). No differences were found in long-term pain outcomes or in reports of dyspareunia.

Our systematic evidence review includes six of the eight trials that were included in the Cochrane review and an additional two trials. Overall, the evidence is from a combination of poor, fair, and good trials, but we considered the strength of evidence as Grade II; moreover, it is consistent with the previous Cochrane review. Basically, evidence indicates that polyglycolic acid sutures are associated with less perineal pain, a lesser need for analgesia use, and fewer healing problems in the short-term postpartum. For long-term outcomes, the evidence is consistent that outcomes of the use of polyglycolic acid sutures and chromic catgut do not differ substantially. One trial not in the Cochrane review⁶⁵ did report more perineal pain and dyspareunia in the polyglycolic-acid group at 6 months, an outcome the authors attributed to the slower absorption rate of polyglycolic-acid sutures; however, these results were neither statistically significant nor precise. Overall, the body of evidence for the comparison of polyglycolic-acid sutures versus chromic-catgut sutures suggests that using polyglycolic-acid sutures for perineal repair offers many short-term advantages.

Combined Methods and Materials. Instead of investigating methods and materials separately, two trials^{49, 55} compared entire approaches, combining both materials and methods in a single randomization design. The poor trial⁴⁹ found no differences between the groups; the fair-quality trial⁵⁵ found that women repaired with polyglycolic-acid sutures using a continuous, subcuticular approach suffered less perineal morbidity. This result is consistent with other trials that investigated subcuticular suturing and polyglycolic-acid sutures separately, perhaps reinforcing the notion that this method and suture type are superior to other options available to obstetric clinicians. Overall limitations in this body of evidence and suggestions for future research are provided later in this chapter.

Summary. The heterogeneity of methods and materials used for repair of episiotomy and perineal laceration arises in part from the passage of time and differences in practice across continents. Another set of issues to consider in studying the repair of episiotomy are the economic and geographic differences among clinical practices across the world. The choice of suture material might be restricted in resource-poor settings. If a clinic cannot afford or does not have access to polyglycolic acid, which, as reported in one of the studies,⁶⁶ is more expensive than other absorbable sutures, then the clinic may have to make do with available sutures but perhaps supplement them with a method of repair that can decrease perineal morbidity.

During the time period that this review encompasses, investigators studied three major classes of suture material (nonabsorbable, absorbable, and tissue adhesive) and two subtypes of sutures (treated versus untreated and standard versus rapidly absorbed). These materials were all studied in the presence of different approaches to the method of suturing; therefore, individual effects of the materials themselves cannot be examined. Likewise, the methods of repair were examined in the context of different materials among the studies and within them for different stages of repair. For these reasons, truly determining the effects of a certain method of repair is impossible, because we are unable to tell whether the outcomes are confounded or modified by suture material.

Key Question 4: Episiotomy and Urinary Incontinence, Fecal Incontinence, and Pelvic Floor Defects

The literature that provides evidence about routine episiotomy, continence, and pelvic floor defects and function is limited in several domains. All but two of the cohorts^{34, 75} report outcomes for use of mediolateral episiotomy rather than midline episiotomy. The length and incompleteness of followup limits the usefulness of the data. We aimed to identify publications with followup ranging from years to decades after births to women with known episotomy histories. However, at completion, only five of 16 publications provide data with followup at 1 year or longer;^{33, 68–70, 74}the longest interval was 5 years.⁶⁹ Followup conducted within 6 months of birth, as in 10 percent of the 16 studies, does not reflect full recovery of the pelvic floor from vaginal birth. For that reason, we considered 6 months followup as an intermediate point of comparison rather than an evaluation of final pelvic floor and continence status.

All measures for followup at or beyond a year were self-reported by interview or questionnaire except a single study with perineometry conducted at 12 months.⁶⁸ Neither self-report for many urinary and rectal continence symptoms nor perineometry and physiologic measures have been adequately validated. These measures do not relate directly to physical examination findings or individual functional status. Indeed, urodynamic testing and physical examination have very limited documentation of their ability to predict or classify continence status. These limitations must be kept in mind. The greatest clinical relevance would be to assess continence and pelvic floor deficits among women with known episiotomy histories beginning in their 40s and proceeding through their lifetimes. None of the identified studies provided such data.

Given these limitations of timing and methods of followup, these prospective studies did not identify improvements in continence for urine or stool or in pelvic floor muscle function among women who had had episiotomy compared to those who had not. This finding includes comparison to women who had spontaneous lacerations of similar severity. Several authors reported decrements in pelvic floor function among women who had episiotomy. Only a single study, using multivariable models, found that episiotomy was an independent predictor of urinary continence.⁶⁹ In the majority of other studies using multivariate models, adjusting for factors such as parity, neonatal weight, and length of second-stage labor revealed that episiotomy was not an independent risk factor for incontinence. Taken in total, this literature, predominantly of fair to poor quality, does not support use of episiotomy for the purpose of preventing pelvic floor defects, urinary incontinence, or incontinence of stool or flatus. Table 11 shows this as graded strength of evidence Grade III.

In summary, these prospective studies are limited because they do not follow women long enough to detect disease occurrence. At present, the assumption that intermediate variables such as pelvic muscle strength measured by perineometry, urodynamic test results, or early reports of symptoms can predict later disease has not been validated. Prospective evaluation only during the months after birth when the pelvic floor is still in a recovery and stabilization period may be misleading. Conclusions about whether episiotomy prevents or increases risk for incontinence and prolapse later in adult life cannot be reached from currently available randomized and cohort studies.

Key Question 5: Episiotomy and Future Sexual Function

The studies addressing this question need to be considered in three groups: the six studies that virtually exclusively evaluate effects of mediolateral incision,^{23, 33, 71, 72, 80, 81} three that evaluate midline (median) incision,^{29, 34, 38} and one that compared the two incision types.³⁶ From just the clinical trials of episiotomy strategy—liberal versus restrictive—one trial addresses each type of incision, and one directly compares the two incision types. None finds substantive differences in sexual function. Overall, this body of literature supports a conclusion that perineal trauma is associated with probability of pain with intercourse in a dose-response fashion such that greater perineal injury is associated with greater probability of pain.

The quality of the evidence (to which we assigned Grade III) to assess this question is limited for reflecting on the consequences of episiotomy. Both the clinical trials and the prospective cohorts assess overly simplified measures of sexual function. Definitions and specification of approaches to measurement are insufficient in many studies to assure accurate interpretation of findings. Validated instruments intended to assess nuances of sexual function such as type of pain, location of pain, severity of pain, orgasm, lubrication, and libido have not been deployed in the published research to assess prospectively the influence of perineal trauma in childbirth on future sexual function. More complex measures would need to be used to understand properly relationships between perineal trauma and future sexual function. Specific factors such as prior sexual function and current libido, in addition to factors such as duration of second-stage labor, size of infant, and lactation status need to be incorporated into multivariable models to generate more-informative and less-biased estimates of the long-term effects of episiotomy in this area. With these caveats, the evidence does not suggest that episotomy results in improved sexual function outcomes.

Deficiencies in The Literature

Our systematic review should be interpreted in the context of several limitations. First, as with all systematic reviews, its findings depend on the predefined approach to searching the literature and on the quality of the published literature identified. The limitations of the available studies (see Chapter 3) include the following:

• age of the data (trials from the 1980s and early 1990s were conducted in an era when background rate of episiotomy was higher);

• insufficient size of most trials for assessing clinically relevant endpoints (third- and fourth-degree lacerations, long-term incontinence);

• inadequate specification of a priori primary and secondary outcomes with few references to power calculations to determine required study size;

• infrequent use of multivariate modeling to account for shortcomings of randomization or need for stratification within RCTs or potential confounders in prospective cohort studies;

• infrequent use of masking of the assessor for outcomes;

• use of a wide variety of measures and timepoints for maternal postpartum outcomes, making comparisons among studies difficult;

• rare use of validated outcome measures;

• limited reporting of precise definitions of self-reported outcomes, particularly for pain, sexual function, and incontinence; and

• inconsistent reporting of appropriate statistical measures (i.e., use of P values without measures of magnitude or confidence intervals), making it difficult to determine if null findings represent lack of effect or limitations in power.

An additional limitation of the literature on KQ 1 results from the nature of the intervention. In essence, KQ 1 reviews studies of clinician behavior when asked to implement different policies for episitomy use. Differences in episiotomy rates observed in groups assigned to routine (i.e., liberal) use vary from a low of 23 percent²⁹ to a high of 52 percent,³⁰ suggesting that clinician behavior is not easily modified, even in the context of an RCT. Inconsistencies in the way that clinicians define and interpret routine use and restricted use of episiotomy within and across trials may temper differences between protocol groups. Additionally, violations of protocol can invalidate initial power calculations and lead to results that cannot be interpreted.

Limitations to Our Review Procedures

Our review process also had some limitations. Because of time and resource constraints, we did not conduct dual, independent, blinded review of articles for inclusion or abstraction of information into evidence tables. Instead, one reviewer performed the initial review, and a second reviewer examined that input and recommended changes. Differences were reconciled between the two reviewers. We used dual review for grading the quality of individual articles, allowing us to evaluate rigorously systematic bias in these assessments.

SECTION 1: ABSTRACTION IDENTIFIERS

1. Article number: _______________________________

2. Abstractor Name: _______________________________

3. Date of abstraction: __ / __ / ___

4. Name of second abstractor/reviewer: _______________________________

5. Date of review: __ / __ / ___

SECTION 2: ARTICLE IDENTIFIERS

1. Year Published: _______________________________

2. Surname of first author: _______________________________

SECTION 3: HEALTH & GEOGRAPHIC SETTINGS

1. Where was the study conducted? (List all countries)

   _____________________________________________

   _____________________________________________

   _____________________________________________

   _____________________________________________

SECTION 4: FUNDING SOURCE

SECTION 5: OBJECTIVE OF THE STUDY

1. _______________________________________________________________________________

   __________________________________________________________________________________

   __________________________________________________________________________________

   __________________________________________________________________________________

   __________________________________________________________________________________

   __________________________________________________________________________________

SECTION 6: INCLUSION/EXCLUSION CRITERIA, STUDY DESIGN, DATA COLLECTION

1. Inclusion Criteria

   __________________________________________________

   __________________________________________________

   __________________________________________________

   __________________________________________________

2. Exclusion Criteria

   __________________________________________________

   __________________________________________________

   __________________________________________________

   __________________________________________________

1. Total number of randomization arms in this study ___________________

2. Description of each randomization arm

   Group 1: _______________________________________________________________________

   Group 2: _______________________________________________________________________

   Group 3. _______________________________________________________________________

3. Describe randomization methods

   _____________________________________________________________________________________

   _____________________________________________________________________________________

   _____________________________________________________________________________________

   _____________________________________________________________________________________

4. Describe blinding/masking methods and how maintained

   _____________________________________________________________________________________

   _____________________________________________________________________________________

   _____________________________________________________________________________________

   _____________________________________________________________________________________

Comments: _____________________________________________________________________________________

______________________________________________________________________________________

______________________________________________________________________________________

SECTION 7: CHARACTERISTICS OF PARTICIPANT

Comments: _______________________________________________________________________________

__________________________________________________________________________________________

__________________________________________________________________________________________

SECTION 8: CHARACTERISTICS OF LABOR & DELIVERY

Comments: _______________________________________________________________________________

_________________________________________________________________________________________

_________________________________________________________________________________________

SECTION 9: CHARACTERISTICS OF REPAIR

Comments: ______________________________________________________________________________

_________________________________________________________________________________________

_________________________________________________________________________________________

SECTION 10: DESCRIPTION OF PARTICIPANTS AVAILABLE @ EACH POINT OF FOLLOWUP

Please describe the number of participants that contributed data to each timepoint of followup during the study and reasons for missing data if applicable. Please refer to section 7, question #20 on page 4 of this form for number of participants in each group at the time of randomization or enrollment.

Comments: ______________________________________________________________________________

_________________________________________________________________________________________

_________________________________________________________________________________________

_________________________________________________________________________________________

_________________________________________________________________________________________

SECTION 11: OUTCOMES (Record as presented in article, as N or %. Follow percent with %)

*Additional space for outcomes can be found in the section 11 addendum on page 16.

ADDENDUM: SECTION 11

Please use this table for additional information on outcomes from above

PAIN ADDENDUM: SECTION 11, #46 PERINEAL PAIN

Please use this table for additional information on perineal pain

SECTION 12: QUALITY ASSESSMENT

Abstractor's Initials ___ ___

Article #: ________________

Assessment of Quality of Individual Articles for RCT's