Farrell, Scott A. MD; Gilmour, Donna MD; Turnbull, Geoffrey K. MD; Schmidt, Matthias H. MD; Baskett, Thomas F. MD; Flowerdew, Gordon PhD; Fanning, Cora A. BN
In a prospective observational study of primiparous women, we found that flatal incontinence 6 months after vaginal delivery was higher in women who sustained a third- or fourth-degree laceration when compared with women who delivered vaginally without injury to the anal sphincter.1 Repair of the external anal sphincter after an obstetric perineal laceration has not produced ideal anal continence results. High rates of anal incontinence after surgery may be explained by the finding of Sultan et al who reported external anal sphincter defects in 82% of women 7 weeks after primary repair of an anal sphincter tear compared with 33% in women who did not have a sphincter tear.2 They demonstrated that anal incontinence was significantly associated with an external anal sphincter defect. Their findings suggested that traditional end-to-end surgical repair of external anal sphincter defects does not restore normal anatomy or function. Sultan et al conducted a small retrospective case series where overlapping repair was performed in an operating room under regional or general anesthetic in 32 women using polydioxanone suture. At 5 months postpartum, only two (7%) women had flatal incontinence and none had fecal incontinence.3 Because of this work, the investigators recommended the overlapping rather than end-to-end surgical technique be used to repair obstetric injuries of the external anal sphincter.
Three randomized studies examined the effects of overlapping compared with end-to-end repair of the external anal sphincter. Fitzpatrick and Williams found no difference in fecal incontinence rates.4,5 Fernando et al also found no difference in fecal incontinence rates but did find a decrease in fecal urgency with the overlapping repair.6 A recent Cochrane review of the literature by Fernando et al which included only these three studies concluded that the evidence suggested some advantage of the overlapping repair technique but that none of those studies reported on surgeon experience.7 The clinical trials conducted to date have been limited in that they included multiparous women, women with partial tears of the external anal sphincter and were unclear as to how their outcome measures were interpreted. We designed the current study to overcome some of the limitations of the previous clinical trials.
The principal objective of this randomized controlled trial was to compare overlapping repair with end-to-end repair of obstetric tears and to investigate which procedure results in a higher rate of flatal incontinence. The secondary objectives were to examine rates of fecal incontinence, rates of external anal sphincter defects by ultrasonography, and anal function using anal manometry.
MATERIALS AND METHODS
This study was conducted in a single tertiary care academic center. Between 2001 and 2007, women who had sustained obstetric third- or fourth-degree lacerations were enrolled in this study. Women with partial tears of the external anal sphincter and multiparous women were excluded from this analysis. The state of the internal anal sphincter was not considered for the inclusion criteria, but any internal anal sphincter defects identified at surgery were repaired. The research protocol was reviewed and approved by the Izaak Walton Killam Health Center institutional review board.
The sample size calculation was based on previous research in our center which found that 47% of primiparous women who sustained a third- or fourth-degree perineal laceration experienced flatal incontinence.8 In Sultan's retrospective series of primary overlapping repair, the rate of flatal incontinence was only 7%.3 To detect the difference between a 50% rate of flatal incontinence and a 25% rate with a power of 80% and an alpha of 0.05 required 60 participants in each study group. A two-tailed test and two-sided confidence interval (CI) for the difference in flatal incontinence rate was planned in reporting the results. Anticipating a dropout rate of approximately 20%, we proposed to enroll a sample size of 150 primiparous women with 75 women in each treatment arm. An interim analysis was planned when the primary outcome would be known for 50% of the patients. To preserve the 0.05 alpha level, we used an O'Brien-Fleming decision rule whereby statistical significance would be declared if either P<.0051 at the interim analysis or P<.0475 at the final analysis was found.8
Among the three studies using anal ultrasonography after primary end-to-end repair of external anal sphincter, Nielsen et al9 found anal ultrasonography defects in 54% of their patients. In Sultan's retrospective series, defects were found in 15% of patients with primary overlapping repair.3 Our study has 80% power to detect a reduction from 50% to 25% in the rate of ultrasound defects of the anal sphincter.
Women were assigned randomly to overlapping or end-to-end external anal sphincter repair using a block size that varied at random (2, 4, or 6) to prevent the ability to predict the group to which the next patient would be assigned. Each randomization required logging into a secure, password-protected Web site accessed through a computer terminal in the birth unit. The allocation sequence was determined by a computer algorithm developed by Dalhousie University Computer Services, who also created the Web site. All study investigators, including the statistician, were blind to the allocation code until the final analysis. The interim analysis was presented by the statistician (with types of surgical repair labeled A and B) to an independent oversight committee whose task was to review any adverse events and instruct the investigators to either stop or continue the study.
Women were consented in two steps. The initial consent and randomization was to one of the repair groups. Surgeons were asked to indicate at the time of repair whether defects in the internal anal sphincter and the anal mucosa were present. A tear that included the internal anal sphincter was classified as 3c and one including the anal mucosa a fourth-degree tear.6 The day after the procedure, participants were approached by our research nurse for consent to the 6-month follow-up.
Both the women and follow-up assessment personnel were blinded to the surgical procedure performed. The women were advised, as part of the consent process, that this blinding was necessary to avoid biasing the results. The healthcare personnel who performed the follow-up evaluations did not have access to the woman's chart for the duration of the study. At the beginning of the study, a grand rounds presentation on the study design and surgical techniques was held. Teaching sessions for the residents and staff were held every 3 months to review the surgical technique. The training was designed to eliminate any confounding due to surgeon inexperience with either method of repair. The principal obstetric researchers made themselves available during the first year of the study to act as preceptors during the repairs.
The surgical repair was performed in layers beginning with a running closure of the anal mucosa with a 3–0 polyglycolic acid (Vicryl, Ethicon, a Johnson & Johnson company, Somerville, NJ) suture. When there was a defect in the internal anal sphincter, it was reapproximated with interrupted sutures of 3–0 polyglyconate (Maxon, Covidien, Dublin, Ireland). The end-to-end repair of the external anal sphincter was achieved with a minimum of two figure of eight sutures of 3–0 polyglyconate (Maxon). We chose this suture over polydioxanone because it has similar durability with superior handling features. The overlapping repair was achieved with of 3–0 polyglyconate (Maxon) using the method described and taught by Sultan.6 One of the obstetric authors (S.A.F.) attended a course in the United Kingdom and was certified in the overlapping repair method and anal ultrasonography. After mobilizing the external anal sphincter, Sultan describes the placement of two to three interrupted sutures approximately 1.5 cm from the edge of the side of the external anal sphincter which is to be overlapped. When tied, these sutures had the effect of pulling one side of the muscle over the top of the other. Additional interrupted sutures were used to fix the overlapped muscle in place. After the repair, surgeons (the person who actually performed the surgery, whether resident or staff person) used a standardized form to record the relevant data for the repairs. Peri-operative intravenously antibiotics were recommended in the form of single doses of 1gm cefazolin and 500 mg metronidazole. Women who were penicillin allergic were given single doses intravenously 450 mg of clindamycin and 100 mg of Gentamicin. Our research nurse subsequently abstracted additional relevant obstetric data from the medical record.
The 6-month follow-up included completion of the incontinence questionnaire. This questionnaire was adapted from the Fecal Incontinence Quality of Life scale10 to make it a more disease- and age-specific instrument.11 The questions used to elicit the outcomes of flatal and fecal incontinence required a yes/no answer. For flatal incontinence, we asked “During the past 4 weeks, have you experienced involuntary passing of flatus from the rectum (passing gas when you didn't want to)?” For fecal incontinence, we asked, “Over the past 4 weeks have you experienced involuntary leaking or soiling of stool (feces) from the rectum?” Additional questions were designed to determine the frequency and quantity of flatal/fecal incontinence and to determine whether the condition had remained the same, deteriorated or improved since delivery. A 45-item quality of life instrument11 specific to anal incontinence was also administered to those women who indicated that they had experienced anal incontinence.
Participants were asked to indicate whether they had had to see a physician for problems since their surgical repair; whether they were currently breastfeeding; and whether they were performing Kegel's exercises. A four-question pain index was used to enquire about pain at the repair site and dyspareunia. Finally, women were asked to indicate whether they were experiencing feelings of sadness, discouragement or hopelessness during the month before answering the questionnaire.
At the 6-month follow-up visit, in addition to completing the standardized questionnaire, women underwent anal ultrasound assessment and anal manometry. The individuals performing these assessments were blinded to the questionnaire results and treatment group of the patient. Anal ultrasonography was performed using a B&K Medical scanner (B&K Medical, Sandhoften, Sweden) with 1850 axial type endoscopic probe with a 10 MHz transducer. The ultrasonography was performed by a radiologist (M.H.S.) or by S.A.F.
During the anal scan, a standardized form was used to record the presence or absence of defects in the internal anal sphincter at the deep and superficial levels and defects of the external anal sphincter at the deep, superficial and subcutaneous levels. At the two deeper levels, the thickness of the internal and external sphincters was measured at four locations corresponding to 12, 3, 6 and 9 o'clock using the electronic calipers of the ultrasound machine. At the subcutaneous level, the thickness of the external anal sphincter alone was measured in the same manner. At the conclusion of the ultrasound examination, the ultrasonographer was asked to give his or her global impression as to the presence or absence of a defect in the external anal sphincter and the internal anal sphincter and to indicate whether he or she had any suspicion as to the type of repair that had been performed.
The anal manometry was performed by coauthor G.K.T., a gastroenterologist with extensive experience in this investigative technique. At the time of anal manometry, a digital rectal examination was performed and the examiner was asked to assess the sphincter anatomy, the resting tone of the sphincter, and the nature of the voluntary contraction. Anal pressures were measured at four positions (4 cm, 3 cm, 2 cm, and 1 cm from the anal verge) and in four quadrants at each position (posterior, left, right, anterior). At each position the maximum resting pressure, the maximum sphincter voluntary contraction pressure, and the pressure with voluntary sphincter contraction for 10 seconds (sustained contraction tone) were measured. The internal anal sphincter length was measured as well. The anal manometry studies were conducted using eight-channel catheters with the Laborie anorectal manometry system (Laborie Medical Technologies, Inc., Montreal, Canada). At completion of the anal manometry, the operator was asked to provide an overall impression of resting sphincter tone; the maximum voluntary contraction tone; and the sustained contraction tone.
Women were reminded by phone of their follow-up appointment. Women who missed their 6-month follow-up were contacted by phone on up to two occasions to rebook their appointments. Those who were unable to attend or declined attendance were sent the incontinence questionnaire by mail and asked to complete it. Subsequent follow-up with the same 6 month questionnaire was arranged annually for 3 years and will be reported separately.
Statistical analysis was performed using SPSS 16 (SPSS, Inc., an IBM company, Chicago, IL) and SAS 9.2 (SAS Institute, Inc., Cary, NC). Specific methods of data analysis to compare baseline characteristics and outcomes for the two surgical repair types consisted of χ2 tests for categorical variables, t tests for continuous variables, and logistic regression to adjust for covariates. The Mantel-Haenszel χ2 test for trend was used to compare surgeons' levels of previous experience with the type of repair used.
Figure 1 is the CONSORT diagram outlining the enrollment and follow-up of women. Six hundred and seventy-one women were assessed for eligibility. Four hundred and ninety-seven women were excluded: 416 did not meet inclusion criteria (most commonly because they did not have a complete third-degree tear), 28 refused participation and 53 were not included for other reasons which included physician decisions concerning other contraindications. A total of 174 were assigned randomly, 86 to an end-to-end repair and 88 to an overlapping repair. Eleven women were excluded after randomization for incomplete sphincter tears. Fifteen women either declined follow-up or were lost to follow-up. Twenty-five multiparous women were excluded from the analysis of this sample of exclusively primiparous women. One hundred and twenty-three women completed the 6-month questionnaire, 62 in the end-to-end group and 61 in the overlapping group. Seventy-six women underwent anal manometry and 74 underwent anal ultrasound studies, 72 of the anal ultrasound assessments were satisfactory for analysis.
Table 1 examines the maternal and newborn characteristics of the study groups. There were no significant differences in these characteristics. Table 2 shows the baseline delivery characteristics of the study groups. All deliveries were of singletons and all were delivered in vertex presentation. Eighty-nine percent of fetuses were delivered in the occiput anterior position, 7% were occiput posterior and 4% were occiput transverse. The majority of women, 99 (80%), had epidural anesthesia during labor and delivery. There were no significant differences in the delivery characteristics.
In the 123 women who completed the 6-month follow-up, 77 (63%) repairs were performed by residents under direct staff supervision, 35 (29%) by obstetricians, eight (6%) by urogynecologists, and three (2%) by fellows. Seventy-eight procedures (63%) were performed with one assistant, 44 (36%) with two or more assistants, and one by a lone surgeon. One of the principal authors was present for 25 (20%) of the procedures; there was no significant difference in principal-author attendance between the two types of surgical repair. There was a significant difference between the two surgical groups in terms of surgeon's experience with the procedure (Table 3). In 37 (60%) of the end-to-end repairs, the surgeon had done 10 or fewer repairs of that type compared with 48 (79%) overlapping repairs (Mantel-Haenszel test for trend, P<.003). Twice as many surgeons, eight compared with four, were performing an overlapping repair for the first time compared with the end-to-end repair.
Ninety-nine participants (80%) received antibiotic prophylaxis. This did not vary between groups. One woman in the end-to-end group had pus and inflammation in the incision site. One woman in the overlapping group experienced a dehiscence. Only one woman had symptoms and signs of an infection in the immediate postpartum period. Twenty-three (19%) visited their doctor with concerns about their incisions but only one woman was readmitted to hospital. Two women in the overlapping group experienced rectovaginal fistulae which were subsequently repaired.
The overall flatal incontinence rate at 6 months in the entire study group was 48%. Thirty-seven (61%) of the women who had an overlapping repair complained of flatal incontinence at 6 months, and 24 (39%) of the women who had an end-to-end repair complained of this problem. Risk of flatal incontinence was significantly higher in women who underwent overlapping repair (odds ratio [OR] 2.44, CI 1.18–5.04 P=.015). The overall rate of fecal incontinence at 6 months was 11%. Nine (15%) women who had an overlapping repair complained of fecal incontinence and 5 (8%) women who had an end-to-end repair complained of fecal incontinence. The difference between these rates was not statistically significant (OR 1.97, CI 0.62–6.27 P=.243). Obstetric and demographic factors examined for their effect on the incidence of flatal and fecal incontinence included method of delivery, episiotomy, condition of the internal anal sphincter at the time of repair, type of tear, location of surgical repair (birth unit compared with operating room), the duration of labor, the woman's age, and the birth weight and head circumference of the neonate. On univariable analysis none of these factors was found to affect the rate of flatal or fecal incontinence. Stepwise logistic regression, both forward and backward methods, was conducted to investigate the effect of the type of repair adjusting for intactness of the internal anal sphincter, intactness of rectal mucosa, presence of a specialist, maternal age and surgeon experience. In this analysis flatal incontinence risk was significantly increased with overlapping repair of the external anal sphincter (OR 3.22, CI 1.45–7.15, P=.004).
An inverse relationship between surgeon experience and flatal incontinence was found. Patients who had procedures performed by surgeons with less experience had lower rates of flatal incontinence (OR 0.65 per unit increase in experience category, CI 0.43–0.97, P=.035). Further examination found that the surgical experience effect occurred between 0 previous repairs (flatal incontinence rate 2/12) and more than 0 previous repairs (flatal incontinence rate 59/111): there was no difference between surgeons who had experience with 1 to 5 compared with 6 to 10 compared with more than 10 previous repairs. Presence of expert assistance, type of surgeon and number of assistants did not explain this finding. There were no significant predictors for fecal incontinence.
Among women who complained of anal incontinence, the frequency of flatal incontinence was 34% for every day or almost every day, 45% for 2 to 5 days per week, and 21% for 1 day per week or less. Frequency of fecal incontinence was 14% for every day or almost every day, 29% for 2 to 3 days per week, and 57% for 1 day per week or less. For the quantity of flatal incontinence, 75% of women complained of minimal to moderate and twenty-five percent had moderate to large. The quantity of fecal incontinence was minimal to moderate in 93% of women and large in only 7% of women. Neither the quantity nor the frequency of anal incontinence differed on the basis of surgical repair.
The effect of the repair type on rates of anal sphincter defects was examined (Table 4).
Seventy-two women had satisfactory anal ultrasonography performed at 6 months postpartum. In 30 of 72 (42%) there was a defect in the internal anal sphincter, 14 in the overlapping group and 16 in the end-to-end group. At the time of delivery, 27 (90%) of these women were noted to have defects in the internal anal sphincter. This difference in rates of internal anal sphincter defects was not significant (OR 0.72, CI 0.28–1.85). In 41 (57%) there was a defect in the external anal sphincter, 23 (62%) in the overlapping group and 18 (53%) in the end-to-end group. This difference in the rates of external anal sphincter defects was not significant (OR 1.46, CI 0.56–3.76). A finding of a defect in both sphincters was not increased in either group (OR 1.00, CI 0.37–2.72).
Table 5 shows the effect of the presence or absence of a defect in the internal or external anal sphincter on flatal and fecal incontinence rates. Although an isolated disruption of either the internal anal sphincter or the external anal sphincter did not affect either flatal or fecal incontinence rates, when a combined disruption of both internal anal sphincter and external anal sphincter was present, it was associated with a higher rate of fecal incontinence.
Examination of the data concerning the ultrasonographer's impressions of the repair type showed that, in 35 (49%) women, the ultrasonographer could not guess the type of repair. In 33 (46%), the ultrasonographer's impression was that the patient had undergone an overlapping repair. Only 17 of these 33 women actually had undergone an overlapping repair, yielding a 52% accuracy rate. In four cases, the ultrasonographers thought that women had undergone an end-to-end repair. Only two actually had undergone this procedure, yielding a 50% accuracy rate.
The effect of the method of repair on manometry readings was examined and, although the pressures were generally higher in the end-to-end group, these differences did not reach statistical significance. The physical examination and summary measures of anal manometry were examined to determine whether any of these findings had an effect on anal incontinence. Sphincter anatomy, resting sphincter tone, voluntary contraction, and maximum voluntary contraction did not affect either flatal or fecal incontinence rates. These measures of anal sphincter function did not differ significantly between surgical groups. If a sustained voluntary contraction was weak or absent, this was associated with higher rates of flatal incontinence (OR 3.5, CI 1.1–11.2). Detailed analyses of the correlation between the ultrasound and anal manometry measurements will be published in a separate article.
Traditional obstetric practice has been to repair external anal sphincter defects using an end-to-end technique. The publication by Sultan which reported a high rate of external anal sphincter defects and associated anal incontinence symptoms after repair using the traditional end-to-end technique prompted concern among clinicians.12 When a subsequent small retrospective study suggested that the rate of postoperative defects was lower and anal function improved after an overlapping repair it was suggested that the standard of care should be changed to overlapping repair.3
Three trials comparing the overlapping to end-to-end repair techniques have been published. A Cochrane review of the three trials found no difference in flatal or faecal incontinence rates between surgical groups but did find a lower rate of faecal urgency and a lower overall anal incontinence score in the overlapping group. It also found no difference in quality of life between groups. All of the foregoing trials had limitations that affected the authors' ability to reliably determine which procedure is best for women. The principal limitations were: the inclusion of multiparous women and women with partial tears of the external anal sphincter; and lack of clarity as to how their outcome measures were interpreted.
Because multiparous women may have already sustained a subclinical impairment of their anal continence mechanism during a “normal” vaginal delivery their inclusion in a clinical trial assessing outcomes of surgical repair of external anal sphincter tears may reduce the likelihood of finding a difference between the groups.13 Of the four randomized prospective studies performed to date, only Fitzpatrick4 and the current study excluded multiparous women. Both Williams and Fernando enrolled multiparous women.5,6 Although these two studies enrolled equal proportions of multiparous women in their study arms, the presence of multiparous women may have enlarged the group of women symptomatic for anal incontinence and blunted the effect of the surgical technique.
We enrolled only women who had sustained a complete third-degree or fourth-degree tear of the external anal sphincter because a complete tear is necessary to the performance of an overlapping repair and women with incomplete tears may have less severe anal incontinence symptoms. Both Fitzpatrick and Williams included women with incomplete third-degree tears in their studies without providing any rationale for their inclusion.4,5 By including women with incomplete tears, these authors enrolled patients who would be less likely to show external anal sphincter defects on anal ultrasonography and also whose sphincter function might be less effected.
We chose flatal incontinence as our primary outcome measure because it is more prevalent than fecal incontinence and has a significant effect on quality of life.1 The three previous clinical trials focused on the much less common symptom of fecal incontinence and relied on scores from questionnaires to make the diagnosis. Both Fitzpatrick and Fernando used modifications of the Wexner14 scale to determine continence status but it is unclear how the authors used the result of the Wexner score to determine the presence or absence of fecal incontinence.4,6 Williams' used the St. Mark's bowel questionnaire but did not explain how this questionnaire was interpreted.5 How these questionnaires were interpreted may have lead to the failure of these researchers to find a significant difference in anal incontinence rates.
In our study, the surgical protocol called for the identification and repair of the internal anal sphincter when torn. Despite this emphasis on repair of the internal anal sphincter, 27 (90%) women identified at delivery with a defect were found to have a defect on anal ultrasonography at 6 months. The surgical identification and repair of this muscle is the most unfamiliar and difficult part of the repair of a 3c tear for an obstetrician. The study surgical protocol emphasized the appearance of the internal anal sphincter as smooth white sheet of tissue with a thickness of 2 mm to 3 mm in close approximation with the anal mucosa. It often retracts laterally and superiorly and may not be incorporated into the repair if it is not specifically identified. It is not possible from the study protocol to know whether these defects found on anal ultrasonography would have been larger had the internal anal sphincter not been repaired at the time of delivery. The internal anal sphincter contributes significantly to the resting anal sphincter pressure, which is critical to anal continence. Research has found that a persistent internal anal sphincter defect after repair of a third- or fourth-degree tear is associated with fecal incontinence.15 The current study did not find this association with an isolated internal anal sphincter defect but did find a higher risk for fecal incontinence when a combined defect of both the external anal sphincter and internal anal sphincter was found on ultrasonography. Logically, the inclusion of this muscle in the repair of a 3c or fourth-degree tear should result in a better outcome and continued efforts should be made to train obstetricians to seek and repair a defect in the internal anal sphincter.
The Cochrane review highlighted the need to assess the effect of the surgeon's experience on outcomes.7 During the current study, repairs were performed by residents, the majority of whom were supervised by staff obstetricians and by obstetricians. Not surprisingly, the surgeons in our study had more experience performing the end-to-end repair. Our finding that surgeon inexperience was beneficial is probably a statistical anomaly. This benefit was only significant when surgeons who had no experience with a procedure were compared with all others. Put another way, this finding confirms that greater experience with either surgical procedure did not improve outcomes. As well, other factors which theoretically should increase the likelihood that a procedure is done well such as the presence of expert assistance and more surgical assistants did not produce better results. These findings support the conclusion that it is the surgical procedure rather than the surgeon that affects anal incontinence rates.
The findings of this study suggest that anal incontinence that follows obstetric injury and surgical repair of the external anal sphincter cannot be explained by isolated structural defects in the sphincter muscles. Most likely concomitantly damage to the innervation of the muscle occurs at the time of sphincter injury and results in impaired muscle function.13 The correlation between a combined defect of the sphincters and fecal incontinence and the association between compromised overall resting tone of the sphincter and flatal incontinence support this dual etiology as the explanation of anal incontinence in these women.
This is the first trial to enroll exclusively primiparous women with complete tears of the external anal sphincter. The results of this report confound previous studies that found either a benefit to an overlapping repair or no difference in the two procedures. Our study finds that the overlapping repair is more likely to result in anal incontinence than the more traditional end-to-end repair. We also found higher rates of external anal sphincter defects in the overlapping group, and although this difference did not reach statistical significance, it does not support the previously reported superiority of the overlapping repair at achieving anatomic normality of the sphincter. The theory that structural integrity is paramount to anal continence fails to take into account the effect of postoperative innervation and sphincter function on the outcome of the surgical procedure. The overlapping repair of the external anal sphincter requires, by necessity, greater mobilization of the muscle to permit successful overlapping of the muscle. It is possible that this dissection causes greater denervation or postoperative scarring which could affect sphincter function. The fact that defects in the internal anal sphincter contributed to the risk of fecal incontinence highlights the importance of including a repair of a defect in the internal anal sphincter as an integral part of the surgical approach.
We conclude that obstetricians should use the more traditional and familiar surgical technique of end-to-end repair for third- or fourth-degree tears of the external anal sphincter sustained at obstetric delivery and should carefully seek and include repair of internal anal sphincter defects.
© 2010 by The American College of Obstetricians and Gynecologists.