Overlapping sphincteroplasty is a well-established

Transcription

1 ORIGINAL CONTRIBUTION Measure of the Voluntary Contraction of the Puborectal Sling as a Predictor of Successful Sphincter Repair in the Treatment of Anal Incontinence Guillaume Zufferey, M.D. 1 & Thomas Perneger, M.D. 2 & Joan Robert-Yap, M.D. 1 Raphaël Rubay, M.D. 1 & Byadran Lkhagvabayar, M.D. 1 & Bruno Roche, M.D. 1 1 Unit of Proctology, Department of Visceral Surgery, University Hospitals of Geneva, Geneva, Switzerland 2 Department of Clinical Research, University Hospitals of Geneva, Geneva, Switzerland PURPOSE: Overlapping sphincteroplasty is the surgery of choice for incontinent patients with an anterior defect after childbirth. Numerous predictive factors have been proposed, but no preoperative variables have been successfully shown to be reproducible. The purpose of this study was to assess the prognostic value of voluntary contraction of the puborectal sling before sphincter repair for anal incontinence. METHODS: This prospective study evaluated 109 consecutive women who underwent surgery for postobstetric anal incontinence. Voluntary contraction of the puborectal sling was measured by perineal ultrasound before the surgery. Severity of anal incontinence was evaluated preoperatively and postoperatively with the Miller Incontinence Score (total incontinence = 18, complete continence = 0). RESULTS: The proportion of patients with scores e3 was 16.7 percent when the preoperative voluntary contraction of the puborectal sling was e4 mm, 48.1 percent when it was 4.1 to 8 mm, and 98.7 percent when it was 98 mm (P G 0.001). Using e8 mm to define abnormal shortening, the sensitivity of the test was 0.95 (95 percent confidence interval, 0.75Y1.00) and specificity was 0.84 (95 percent confidence interval, 0.75Y0.91). CONCLUSION: A preoperative voluntary contraction of the puborectal sling 98 mm convincingly discriminates between patients with a good functional outcome and Address of correspondence: Bruno Roche, M.D., Unit of Proctology, Department of Visceral Surgery, University Hospitals of Geneva, 1211, Geneva 14, Switzerland. bruno.roche@hcuge.ch Dis Colon Rectum 2009; 52: 704Y710 DOI: /DCR.0b013e31819d46a6 BThe ASCRS 2009 those with an unsatisfactory outcome after sphincter repair for postobstetric anal incontinence. KEY WORDS: Fecal incontinence; Outcome assessment; Ultrasonography; Anal canal; Sphincter repair. Overlapping sphincteroplasty is a well-established procedure for repair of an acquired rupture of the anal sphincter, a common complication of childbirth. 1 Whether the procedure will be successful is not easy to predict. Several predictive factors for postoperative continence have been proposed, 2 but improvement of resting 3 and squeeze pressures 4 as measured by preoperative and postoperative anal manometry and length of the postoperative anal canal 3Y5 are the only recognized prognostic variables, and both are based on postoperative analysis. Until now no prognostic factors based on preoperative measurements have been identified, and this clinical deficiency makes it difficult to inform patients adequately about their chances for functional improvement. Pudendal nerve tearing can occur during delivery 6,7 and is implicated as the main cause of a hypokinetic perineum. Clinical intuition suggests that diminished mobility of the perineum or the puborectal muscle seems to be associated with worse results after sphincteroplasty, but no reproducible method of measuring the preoperative mobility of the perineum has been described. Recent studies demonstrate that pelvic floor disorders and the mobility of the perineum can be assessed by dynamic perineal sonography. 8Y10 This technique allows easy identification of the puborectal muscle and a reliable assessment of its mobility. 11 Nevertheless, perineal sonography is not widely used to assess patients before surgical sphincter repair, possibly because its predictive ability is not well documented. 704 DISEASES OF THE COLON & RECTUM VOLUME 52: 4 (2009)

2 Diseases of the Colon & Rectum Volume 52: 4 (2009) 705 The purpose of this study was to determine whether voluntary contraction of the puborectal sling (PRS) predicts functional outcome in patients who undergo corrective surgery for anal incontinence secondary to obstetric trauma. METHODS Study Design, Sample, and Setting We conducted a prospective study of a sample of consecutive patients with anal incontinence and documented rupture of the anal sphincter who underwent an overlapping sphincter repair between 1999 and 2004 at the University Hospitals of Geneva (Geneva, Switzerland). The University Hospitals of Geneva is a teaching hospital system of 2,200 beds and the only public hospital serving the local population. Because this observational study was based on data collected as part of routine patient assessments, it was exempted from formal review by the Hospital Research Ethics Committee. All sphincteroplasties were performed by the same experienced senior surgeon (BR). All patients with sphincter rupture were surgically corrected independent of whether a sphincter contraction was demonstrated or not, as is generally proposed in the literature. 12Y14 The senior surgeon (BR) also performed all perineal sonographies while blinded to the results of the clinical assessment, in particular the Miller score. This surgeon has 20 years of experience in endoanal and perineal ultrasonography. An independent clinical research physician (GZ) collected the continence data and the Miller score before and after surgery. Surgical Intervention All patients received perioperative intravenous antibiotics. No mechanical bowel preparation was prescribed. Sphincter repair was performed using the standard overlapping technique. With a patient placed in the lithotomy position, a perineal curvilinear incision was made in the medial anovulvar line. The disrupted sphincter was identified and gently mobilized without division of the internal and external elements, thus allowing a strong, maximal overlap. The muscles were overlapped with 6 to 8 stitches of long-term absorbable sutures. Special attention was given to constructing a large overlap of the distal portion of the sphincter, creating a repair that extended the length of the sphincter, thus a strong repair. The operative wound was closed mostly vertically in several planes, and no drainage was used. An enema was routinely given immediately before the first defecation. The patients were hospitalized for a short period, generally 3 days. Study Variables The main independent variable was the amplitude of voluntary contraction of the PRS, measured by sonography before surgery. During the preoperative visit, every patient with sphincter rupture underwent a clinical examination and standard endoanal echography. A complementary perineal ultrasound with a plane or curved linear probe of 3.5 MHz was performed, and voluntary contraction of the PRS was measured. The puborectal muscle forms a sling running around the posterior anal canal (Fig. 1A), forming the anorectal angle. With the patient lying in supine position, the probe is placed on the perineum, FIGURE 1. View of the perineum, centered on the puborectal and sphincter muscle in the sagittal plane with the same orientation as used in echographic assessments. A. Representation of the anatomy. B. Echogram. a = internal sphincter; b = external sphincter; c = puborectal muscle; d = anal canal; e = pubic bone; f = ultrasound probe; l = inner border of the external sphincter and the puborectal sling.

3 706 ZUFFEREY ET AL: NEW PREDICTIVE FACTOR FOR SPHINCTER REPAIR FIGURE 2. Sagittal view of the perineum on echography before and after voluntary contraction of the puborectal sling. The white lines indicate the inner border of the puborectal sling. The x markers show the tip of the puborectal muscle. A. Before voluntary contraction. B. After voluntary contraction. C. Anterior displacement of the puborectal muscle. allowing for a sagittal view of the anovaginal bifurcation. The puborectal muscle can be easily recognized in the sagittal plane as a hyperechogenic structure located posteriorly in the superior part of the anal canal. It lies immediately behind the hypoechogenic line of the rectal wall and the internal sphincter, and is prolonged distally by the external sphincter, thus forming the anorectal angle (Fig. 1B). Voluntary contraction of the perineum and the PRS leads to a shortening of the PRS. The anal canal moves in the direction of the pubic bone and diminishes the anorectal angle. In contrast, relaxation of the PRS, which occurs while attempting defecation, lengthens the PRS and opens the anorectal angle. Contraction and relaxation of the PRS can be measured with sonography. After marking a reference point in the middle of the internal border of the PRS in the sagittal view (Fig. 2A), the displacement of the reference point during contraction and relaxation can be followed in a linear direction and measured in millimeters after the image is captured (Figs. 2B and C). The outcome variable was functional outcome at 3 months, evaluated by the Miller score. This widely used score, proposed by Miller et al. 15 in 1988, is used in our hospital system to classify anal incontinence. The score is based exclusively on the frequency of incontinence of solid stool, liquid stool, and gas. A completely incontinent patient has a Miller score of 18, whereas a continent patient has a score of 0 (Table 1). The Miller score was assessed by the same physician preoperatively and at the three-month postoperative visit. Additional variables were the demographic characteristics of the patients. Statistical Analysis We compared the preoperative and postoperative Miller scores using a paired Wilcoxon test. Then we explored the associations between voluntary contraction and the preoperative Miller score, postoperative Miller score, and the difference between the two (functional improvement) using scatterplots, nonparametric locally weighted regression, and analysis of variance. We also dichotomized the postoperative Miller score as asymptomatic (score 0) vs. symptomatic (score 1Y18) and as asymptomatic or flatus incontinent (score 0Y3) vs. incontinence to liquids and/or solids (score 4Y18). On the basis of these analyses, we categorized puborectal voluntary contraction as e4 mm, 4 to 8 mm, 8 to 12 mm, 12 to 16 mm, or 916 mm, and compared these subgroups in terms of their mean Miller scores (using ANOVA) and the proportion with a successful outcome, as reflected by the dichotomous postoperative Miller scores. Based on this analysis, we dichotomized the puborectal voluntary contraction as negative (i.e., abnormal, or e8 mm) vs. positive (i.e., normal, 98 mm). We computed the sensitivity, specificity, and positive and negative predictive values for this test, as well as exact 95 percent confidence intervals for these statistics. To capture the global predictive ability of the puborectal voluntary contraction, we constructed receiver operating characteristic (ROC) curves for this variable against the two dichotomous outcomes (postoperative Miller score of 0 vs. 1Y18 or e3 vs. 4Y18) and computed areas under the ROC curve (AUCs). The AUCs range from 0.5, for a test that does not discriminate at all between good and poor outcomes, to 1.0, for a test that discriminates perfectly. Specifically, the AUC can be interpreted as follows: if one patient with a poor outcome and one patient with a good outcome are chosen at random, the AUC is the probability that the patient with TABLE 1. Miller score for anal incontinence Frequency Gaz incontinence Liquid stool incontinence Solid stool incontinence G1/month /month to G1/week 91/week 3 6 9

4 Diseases of the Colon & Rectum Volume 52: 4 (2009) 707 Analyses were conducted using SPSS 15 software (SPSS Inc., Chicago, IL). RESULTS Patient Characteristics One hundred and nine consecutive female patients underwent a sphincteroplasty in our center between January 2001 and November The median patient age was 48.6 (range, 24Y80) years. The median duration of hospitalization was three days. No immediate or late mortality occurred. Early complications were wound leakage (n = 3) and perineal hematoma (n = 2). Symptom Scores The average preoperative Miller score was 13.3 (standard deviation, 4.2; range, 3Y18). None of the patients were asymptomatic, 2 (1.8 percent) of the 109 patients had a score between 1 and 3 (flatus incontinence only), 37 (33.9 percent) had a score between 4 and 9, and 70 (64.2 percent) had a score of 10 or more. In contrast, 3 months after the surgery, the average Miller score was 2.6 (standard deviation, 4.3; range, 0Y18; P G 0.001), 60 (55.0 percent) patients were completely asymptomatic, 29 (26.7 percent) had a score between 1 and 3, 12 (11.0 percent) had a score between 4 and 9, and 8 (7.3 percent) had a score of 10 or more. The mean score improvement was 10.7 (standard deviation, 5.2; range, 0Y18): 6 (5.6 percent) patients did not improve at all, 8 (7.4 percent) improved by 1 to 3 points, 39 (36.1 percent) improved by 4 to 9 points, and 55 (50.9 percent) improved by 10 points or more. FIGURE 3. Associations between anterior shortening of the puborectal sling and preoperative Miller score (A) and postoperative Miller score (B) in 109 women who underwent surgical repair for anal incontinence. Nonparametric regression lines are superimposed. the poor outcome will have a worse (more abnormal) test result. All statistical models were overfitted to the sample from which they were developed and thus would be expected to perform less well in other samples. To address this problem, we performed a cross-validation procedure. 16 The sample was split at random into five subsets. The association between anterior displacement of the PRS and the probability of a successful outcome was modeled in four-fifths of the sample using logistic regression, and the resulting regression coefficient was applied to the remaining fifth. This was done five times with a different subset excluded each time. We then examined whether the probabilities of success obtained from unrelated data (the other four-fifths) were associated with actual success, using a ROC curve. FIGURE 4. Association between anterior shortening of the puborectal sling (categorized) and clinical outcome after surgical repair for anal incontinence. Clinical outcome is categorized as asymptomatic (Miller score = 0, green bars), flatus incontinence (Miller score = 1Y3, blue bars), symptomatic (Miller score = 4Y9, orange bars), and severely symptomatic (Miller score = 10 or higher, red bars).

5 708 ZUFFEREY ET AL: NEW PREDICTIVE FACTOR FOR SPHINCTER REPAIR TABLE 2. Mean (SD) clinical outcome scores in relation to puborectal voluntary contraction Preoperative Miller score Improvement in Miller score Postoperative Miller score Proportion asymptomatic after surgery Proportion with Miller score of 0Y3 after surgery N Preoperative puborectal P = 0.38* P G 0.001* P G 0.001* P G 0.001* P G 0.001* voluntary contraction 0Y4 mm (3.7) 2.7 (3.6) 12.5 (7.0) Y8 mm (4.1) 8.1 (5.5) 5.9 (4.2) Y12 mm (4.4) 12.1 (4.3) 0.9 (1.8) Y16 mm (4.1) 12.2 (4.3) 0.3 (0.8) mm (4.1) 13.7 (4.1) 1.0 (1.4) *Test for linear trend. Associations Between Preoperative Assessments and Clinical Outcomes Voluntary contraction of the PRS, measured preoperatively, was not associated with concurrent symptom scores (Fig. 3A). The correlation between voluntary contraction and preoperative Miller score was weak (j0.07) and statistically nonsignificant (P = 0.47). In contrast, voluntary contraction was markedly associated with postoperative Miller score (Fig. 3B). The correlation coefficient was strong (0.63) and significant (P G 0.001). We reanalyzed the data with voluntary contraction grouped into strata of 4 mm and postoperative Miller score grouped into clinically meaningful categories: asymptomatic (score = 0), flatus incontinence only (score = 1Y3), symptomatic (score = 4Y9), and severely symptomatic (score Q 10). The proportions of asymptomatic patients and of patients with only flatus symptoms increased as the distance of voluntary contraction increased (Fig. 4). These graphic exploratory analyses were confirmed through comparisons of mean Miller scores across categories of voluntary contraction (Table 2). The mean preoperative Miller score was not associated with voluntary contraction, but both improvement in Miller score and postoperative Miller score were significantly associated in these tests. Test Characteristics Voluntary contraction of the PRS, maintained as a continuous variable, discriminated well between patients with a favorable outcome and those with an unfavorable outcome (Table 3). The areas under the ROC curves were 0.80 for the prediction of any symptoms and 0.95 for the prediction of symptoms worse than flatus incontinence (Fig. 5). These numbers changed only minimally after cross-validation (Table 3). On the basis of the previous analyses, e8 mm was selected to define an abnormal voluntary contraction. The sensitivity of this test to identify any symptoms after surgical repair was moderate (0.61), but the specificity was excellent (0.95), meaning that most patients who eventually became asymptomatic were correctly identified by a voluntary contraction 98 mm. The positive predictive value was also 90.90, meaning that among those who have a voluntary contraction of e8 mm, the majority will have symptoms after surgery. The test performed even better if unfavorable outcome was defined as any symptoms worse than flatus TABLE 3. Test characteristics of voluntary contraction of the puborectal sling as a predictor of unfavorable clinical outcome after sphincter repair for anal incontinence Any symptoms (Miller score 9 0) Any symptoms worse than flatus incontinence (Miller score 9 3) Estimate 95% Confidence interval Estimate 95% Confidence interval Continuous measure of voluntary contraction Area under the receiver operating characteristic curve (observed) Y Y0.99 Area under the receiver operating characteristic curve (cross-validated) Y Y0.99 Dichotomized: voluntary contraction 98 mm Sensitivity* Y Y1.00 Specificity y Y Y0.91 Positive predictive value z Y Y0.75 Negative predictive value` Y Y1.00 *Among those with unfavorable outcome, proportion correctly identified by test (e8 mm). y Among those with favorable outcome, proportion correctly identified by test (98 mm). z Among those with normal test (98 mm), proportion with favorable outcome. `Among those with abnormal test (e8 mm), proportion with unfavorable outcome.

6 Diseases of the Colon & Rectum Volume 52: 4 (2009) 709 FIGURE 5. Receiver operating characteristic curve for anterior shortening of the puborectal sling, measured preoperatively, as a predictor of symptoms worse than flatus incontinence after surgical repair. (Table 3). The sensitivity and specificity were both excellent, and the negative predictive value was highly accurate, meaning that among those with a voluntary contraction 98 mm, almost all will become asymptomatic or will have flatus incontinence only. Only the positive predictive value was moderate, at DISCUSSION Measuring the voluntary contraction of the PRS with perineal ultrasound is a helpful tool for predicting the clinical outcome after sphincter repair for anal incontinence among women who sustain an anal sphincter tear during childbirth. The global discriminative power of this test was considerable, with an area under the ROC curve of This means that if one patient with a favorable outcome after surgery (asymptomatic or flatus only) and another with an unfavorable outcome (incontinence worse than flatus) are selected at random, the probability is 95 percent that the former had a greater voluntary contraction than the latter. Such results are infrequent in the research of clinical diagnostic or predictive techniques. However, when the anatomy is considered, the correlation becomes understandable. The puborectal muscle derives its nerve supply from the sacral plexus (S-2YS-4). Branches from the pudendal nerve supply the sphincter ani externus. The corresponding fibers arise from S-2 to S This sensory and motor reflex plays an important role in maintaining continence. Pudendal neuropathy, mainly related to a traumatic delivery, leads to pudendal conduction disorders and incontinence. 17 Because an overlapping sphincter repair is less likely to be successful in patients with pudendal neuropathy, this condition was proposed as a predictive factor for sphincteroplasty by a number of authors. 18Y20 However, a significant correlation between pudendal nerve terminal motor latency and success of anal sphincter repair could not be established. 14,22 This later finding is not surprising given the poor reproducibility of pudendal nerve terminal motor latency. 14 The puborectal muscle and perineum mobility can be intuitively associated. A measure of puborectal muscle mobility can be interpreted as an index of the pudendal nerve function. It has been well demonstrated in the literature that anal squeeze pressures are not predictive of a successful sphincteroplasty. This fact could be placed in opposition to the good predictive value of voluntary contraction of the PRS demonstrated here. Voluntary contraction of the PRS can be analyzed as an identical voluntary squeezing maneuver. But a rupture of the anal sphincter, the muscle fibers of which are orbicular, diminishes or suppresses the generation of squeeze pressure itself. Because of this mechanical evidence, squeeze pressure cannot reflect eventual sphincter nerve damage. The PRS, in contrast, is never involved in a sphincter tear, and its function is a tool for evaluation of perineal innervation. Clinically patients are asked to squeeze the perineum with no foreign body (i.e., no anal balloon) in the anus. They are asked to contract or squeeze the perineum, which is a more natural and easily performed movement than an isolated contraction of the anal sphincter alone. Because this movement is more comfortable and easier to perform than anal squeeze pressures, it is less influenced by voluntary effort. If patient effort played an influential role, the results of this study would have shown less correlation and patients who made poor squeeze efforts would have had very good surgical outcomes, which was not the case. Dynamic perineal sonography is a new modality for recognition of perineal structures and demonstration of pelvic floor disorders. 8Y10 As we found in a preliminary study based on 12 healthy female volunteers investigated by two blinded observers, 11 the interobserver reproducibility for measuring voluntary contraction of the PRS is good. In the same study, every volunteer showed a normal range of voluntary contraction of 97 mm (mean value, 15.5 mm T 2.2 mm). Perineal ultrasound imaging of the anal canal offers a double advantage and is our preferred technique for measuring voluntary contraction of the PRS. First, it provides a dynamic view of the perineum, and second, it does not deform the sphincter structures, contrary to endoanal ultrasound and magnetic resonance imaging. Reproducible reference points for the measurement of voluntary contraction of the PRS were difficult to find. Immobile reference points, such as the pubic bone or tip

7 710 ZUFFEREY ET AL: NEW PREDICTIVE FACTOR FOR SPHINCTER REPAIR of the coccyx, cannot be visualized correctly with ultrasound. However, the puborectal muscle shortens along a straight virtual line, which can easily be followed during contraction. The angle between the PRS line during contraction and the horizontal reference plane of the probe varies slightly according to individual anatomy but does not influence the measurement of absolute displacement of the tip of the puborectal muscle. The strengths of this study include a representative consecutive sample of patients and a complete follow-up. The test we used requires minimal time and is relatively inexpensive to perform compared with magnetic resonance imaging and other imaging tests. After the mandatory endoanal ultrasound is performed, it is easy to use the probe to obtain perineal images in the same session. Patients tolerate the test well because it is quick and painless. Study limitations include a fairly small sample size and the unknown applicability of the results to other settings. In particular, the results obtained may be related to the choice of surgical technique used (i.e., overlapping anterior sphincteroplasty) and to the skills and experience of the surgeon. Finally, measurement of voluntary contraction of the PRS requires ultrasound equipment and appropriate skills. The results of this study suggest that the value of 8 mm discriminates best between a normal and an abnormal voluntary contraction. Almost all patients with a normal preoperative voluntary contraction became asymptomatic or minimally symptomatic after surgical repair, in contrast to less than half of the patients with an abnormal preoperative voluntary contraction. The results of the preoperative assessment of voluntary contraction of the PRS are useful when discussing the best course of action with the patient. Those with a good result can be reassured about the high probability of a favorable outcome, and those with a worse result, especially if the voluntary contraction is G4 mm, should be warned that symptoms may not be fully relieved by surgery and that other treatments, such as physiotherapy or sacral nerve stimulation, might subsequently be needed. As the present results show, measurement of voluntary contraction of the PRS is a useful tool for predicting the outcome of sphincter repair in patients with anal incontinence after childbirth, and the results of this assessment are now routinely discussed in every preoperative consultation at our institution. REFERENCES 1. Slade MS, Goldberg SM, Schottler JL, Balcos EG, Christenson CE. Sphincteroplasty for acquired anal incontinence. Dis Colon Rectum 1977;20:33Y5. 2. Baig MK, Wexner SD. Factors predictive of outcome after surgery for faecal incontinence. Br J Surg 2000;87:1316Y Wexner SD, Marchetti F, Jagelman DG. The role of sphincteroplasty for fecal incontinence reevaluated: a prospective physiologic and functional review. Dis Colon Rectum 1991;34:22Y Fleshman JW, Dreznik Z, Fry RD, Kodner IJ. Anal sphincter repair for obstetric injury: manometric evaluation of functional results. Dis Colon Rectum 1991;34:1061Y7. 5. Hool GR, Lieber ML, Church JM. Postoperative anal canal length predicts outcome in patients having sphincter repair for fecal incontinence. Dis Colon Rectum 1999;42:313Y8. 6. Dietz HP. Pelvic floor trauma following vaginal delivery. Curr Opin Obstet Gynecol 2006;18:528Y Weidner AC, Barber MD, Visco AG, Bump RC, Sanders DB. Pelvic muscle electromyography of levator ani and external anal sphincter in nulliparous women and women with pelvic floor dysfunction. Am J Obstet Gynecol 2000;183:1390Y9. 8. Beer-Gabel M, Teshler M, Schechtman E, Zbar AP. Dynamic transperineal ultrasound vs. defecography in patients with evacuatory difficulty: a pilot study. Int J Colorectal Dis 2004;19:60Y7. 9. Beer-Gabel M, Teshler M, Barzilai N, et al. Dynamic transperineal ultrasound in the diagnosis of pelvic floor disorders: pilot study. Dis Colon Rectum 2002;45:239Y Roche B, Deleaval J, Fransioli A, Marti MC. Comparison of transanal and external perineal ultrasonography. Eur Radiol 2001;11:1165Y Fransioli A, Weber B, Cunningham M, Roche B, Marti MC, Hadengue A. Dynamic evaluation of puborectalis muscle function by external perineal sonography. Tech Coloproctol 1996;3:125Y Gilliland R, Altomare DF, Moreira H Jr, Oliveira L, Gilliland JE, Wexner SD. Pudendal neuropathy is predictive of failure following anterior overlapping sphincteroplasty. Dis Colon Rectum 1998;41:1516Y Setti Carraro P, Kamm MA, Nicholls RJ. Long-term results of postanal repair for neurogenic faecal incontinence. Br J Surg 1994;81:140Y Chen AS, Luchtefeld MA, Senagore AJ, Mackeigan JM, Hoyt C. Pudendal nerve latency: does it predict outcome of anal sphincter repair? Dis Colon Rectum 1998;41:1005Y Miller R, Bartolo DC, Locke-Edmunds JC, Mortensen NJ. Prospective study of conservative and operative treatment for faecal incontinence. Br J Surg 1988;75:101Y Altman DG, Royston P. What do we mean by validating a prognostic model? Stat Med 2000;19:453Y Londono-Schimmer EE, Garcia-Duperly R, Nicholls RJ, et al. Overlapping anal sphincter repair for faecal incontinence due to sphincter trauma: five year follow-up functional results. Int J Colorectal Dis 1994;9:110Y Oliveira L, Pfeifer J, Wexner SD. Physiological and clinical outcome of anterior sphincteroplasty. Br J Surg 1996;83:502Y Gilliland R, Altomare DF, Moreira H Jr, et al. Pudendal neuropathy is predictive of failure following anterior overlapping sphincteroplasty. Dis Colon Rectum 1998;41:1516Y Sangwan YP, Coller JA, Barrett RC, et al. Unilateral pudendal neuropathy: impact on outcome of anal sphincter repair. Dis Colon Rectum 1996;39:686Y Altomare DF, Rinaldi M, Petrolino M, et al. Reliability of electrophysiologic anal tests in predicting the outcome of sacral nerve modulation for fecal incontinence. Dis Colon Rectum 2004;47:853Y7.