Approximately 670,000 hysterectomies are performed annually in the United States.1 The overall incidence of ureteral injury is low, estimated at 0.03–2.0% for abdominal hysterectomy,2–5 0.02–0.5% for vaginal hysterectomy,2,3,6 and 0.2–6.0% for laparoscopically assisted vaginal hysterectomy (LAVH).3,7–9 However, given the large numbers of hysterectomies performed annually, this translates into a substantial number of ureteral injuries. Ureteral injuries are of particular concern to the practicing gynecologist because 75% of ureteral injuries occur during gynecologic surgery and 75% of these occur during abdominal operations.10–12
Most ureteral injuries are unsuspected and are diagnosed postoperatively.12–14 Patients with undiagnosed ureteral injuries may present with flank pain, fever, costovertebral angle tenderness, ureterovaginal fistula, ileus, urine peritonitis, pyelonephritis, and, in the rare case of bilateral ureteral injury, anuria. Some ureteral injuries may remain asymptomatic and result in silent renal death. In a study by Conger et al,15 two of 13 ureteral injuries were asymptomatic, and in another study by St. Martin,16 two of eight ureteral injuries were asymptomatic.
The ability to identify ureteral injuries adequately at the time of the initial operation is of paramount importance. Permanent damage can usually be avoided if the diagnosis of the ureteral injury is made at the time of surgery.17 Delay can put the patient at risk of permanent loss of renal function and ureterovaginal fistula. Studies have shown that intraoperative recognition and repair of ureteral injuries, as compared with postoperative diagnosis and delayed repair, decrease postoperative morbidity,12,13,18–21 minimize loss of kidney function and need for nephrectomy, and decrease the incidence of ureterovaginal fistula.10,22,23 Because ureteral injury is often not suspected by the surgeon, therefore delaying diagnosis until the postoperative period, Wiskind and Thompson24 suggested that gynecologists should perform routine cystoscopy at the time of hysterectomy. Intraoperative cystoscopy with intravenous indigo carmine offers rapid confirmation of bilateral ureteral patency. No patients with bilateral ureteral patency at the time of intraoperative cystoscopy were later found to have a ureteral injury. Interestingly, four of the five unsuspected ureteral injuries in the series of Wiskind and Thompson24 were noted after vaginal hysterectomy, a procedure generally associated with a lower rate of ureteral injury.
The objective of this study was to determine the cost-effectiveness of universal cystoscopy for identifying unsuspected ureteral injuries at the time of hysterectomy. Because a clinical trial was not practical, a decision-analysis model was developed to address the question of cost-effectiveness using available data and resources.25
Materials and Methods
We first developed a decision-tree model to define the probability, choices, and potential outcomes associated with the use of routine cystoscopy at the time of total abdominal hysterectomy (TAH) in cases of benign disease (Figure 1). All cost estimates are expressed in 1998 US dollars. At the first decision point, one can either perform a cystoscopy or not. If a cystoscopy is performed, it is associated with a fixed cost estimate of $125 (costs of indigo carmine, sterilization of equipment, and operating room time). The cystoscopy may reveal no bilateral spill with a probability equal to the probability of ureteral injury, given our assumption that cystoscopy is 100% sensitive and specific for ureteral injury. Alternatively, it may reveal bilateral ureteral spill, with a probability of 100% minus the probability of ureteral injury.
A modified version of this model was used to address the cost-effectiveness of universal cystoscopy in cases of total vaginal hysterectomy (TVH) and LAVH. This second model was developed to address the increased cost of converting to a laparotomy for surgical repair of injuries that were diagnosed intraoperatively but were not corrected by the removal of a suture thought to be kinking the ureter. For the vaginal-procedure model, the rate of ureteral injury was varied from 0.2% to 5% to incorporate the low rate of injury associated with TVH and the higher rate of injury associated with LAVH into a single model that accounted for the cost of conversion to laparotomy.
If no bilateral ureteral spill is seen at cystoscopy, then further intraoperative evaluation or repair would begin because it is impossible to predict which ureteral injuries will remain asymptomatic. The absence of bilateral spill of indigo carmine could be the result of a kinking injury or a true injury, estimated as one minus the probability of a kinking injury. The probability of a ureteral kinking injury was estimated at 10% based on expert opinion and consensus of the authors. If a kinking injury is diagnosed, it would be repaired, with the associated costs of intraoperative evaluation and prolonged operative time (estimated as an additional $150). If a true injury occurred, then the model assigns a cost of reimplantation, estimated at $3000. Conversion to laparotomy was assigned a base cost of $2500.
Alternatively, at the first decision point, one could elect not to perform cystoscopy. In this situation, the two possibilities are either a ureteral injury (at the same probability as in the cystoscopy arm of the tree) or no injury. In cases of symptomatic ureteral injuries diagnosed postoperatively, the model includes the probability of requiring a percutaneous nephrostomy, estimated at 50%, with an associated additional cost of $2000. No need for percutaneous nephrostomy was estimated as one minus the probability of requiring a percutaneous nephrostomy. Ureteral injuries diagnosed postoperatively were estimated to present as ureteral obstructions in 85% of patients and as ureterovaginal fistulas in the remaining patients. The cost of ureteral reimplantation at a separate surgical procedure was estimated to be required for all cases of ureteral obstruction diagnosed postoperatively and was assigned an estimated cost of $10,000. The cost of repair of the ureterovaginal fistula, including the cost of ureteroneocystostomy, was calculated at $11,000 to account for the additional cost of fistula evaluation and repair. If cystoscopy has no influence on the rate of unrecognized ureteral injuries, then the test efficacy is 0%. Conversely, if the risk of unrecognized ureteral injury is eliminated, cystoscopy has 100% efficacy. The model was constructed and all calculations were performed with DATA 3.5 (Tree Age, Williamstown, MA), a computer program for performing decision analysis. Copies of the model are available from the authors upon request.
If an unidentified ureteral injury occurs, there is a chance that the injury would remain asymptomatic or that the injury would become symptomatic. A symptomatic ureteral injury would present as either an obstruction or a ureterovaginal fistula, estimated as one minus the probability of ureteral obstruction. We set the probability of an asymptomatic ureteral injury to zero in the base case because of a lack of data on its incidence. In addition, from a purely economic point of view, a silent renal death might never be detected or cause the patient any problems, and thus not incur any costs. By assuming that all injuries would be symptomatic, the model is biased toward detection of injuries, because all injuries will incur costs eventually. We tested the impact of this assumption on our results by varying the incidence of silent death up to 50%.
We performed the analysis from the perspective of a hospital. This means that we did not consider the costs of outpatient visits for symptoms attributable to a ureteral injury, the costs of medications associated with such outpatient management, or the nonmedical costs associated with time lost from work or transportation to health care visits, mainly because of difficulty in obtaining such data. We also did not consider the effects of an undiagnosed ureteral injury at the time of surgery on quality of life, again primarily because of a lack of data. Because all of these considerations would make ureteral injuries more costly, not including them biases the model against universal cystoscopy.
For several reasons, we also did not include the potential costs of malpractice litigation. First, because not every injury results in litigation, we had no data on which to base the frequency of lawsuits. Second, because the methodology for how to estimate the costs of litigation in health-economic analyses is not well defined, the vast majority of published analyses do not include these costs. Third, because a lawsuit would only add to the overall cost of the injury, not including these costs biases the model against universal cystoscopy.
The incidence of ureteral injury was obtained from a review of the literature (Table 1). We performed a MEDLINE search for publications from 1966 to January 2000 by combining the following exploded terms: “gynecologic disease/surgery,” “gynecologic surgical procedures,” “genitalia female/surgery,” “laparoscopy,” “pelvis/surgery,” and “uterus/surgery” with the combination of the exploded term “postoperative complication/surgery” and the key words “iatrogenic” and “injury” and separately with the combination of the following exploded terms: “ureter/injury,” “ureteral obstruction/surgery,” “bladder fistula/surgery,” “urinary fistula/surgery,” and “vesicovaginal fistula/ surgery.” We limited the search to human studies in English. Additional references were obtained from the queried articles' reference lists and review articles. We further limited our systematic literature review to gynecologic operations for benign conditions. Expert opinion was used when data were unavailable in the literature. Because ureteral injuries are relatively rare, only articles with more than 200 patients were included, unless the study was designed to provide routine surveillance of ureteral injury in the form of intraoperative cystoscopy or postoperative intravenous pyelography.
Costs associated with separate admissions for ureteroneocystostomy and repair of ureterovaginal fistula were estimated based on the average Medicare reimbursement for “other kidney and urinary tract diagnoses” Diagnostic Related Groups in the Health Care Financing Administration Medicare Provider Analysis and Review database for 1998 (http://www.hcfa.gov/stats/medpar/ss98d&s.txt). An additional $1000 was allotted for repair of ureterovaginal fistulas above the estimated cost of repair of ureteral obstruction to account for the increased complexity of the surgical procedure. The costs associated with cystoscopy, suture release, percutaneous nephrostomy, reimplantation at the original operation, and conversion to laparotomy were difficult to obtain and were estimated based on Duke University Medical Center data. Probabilities and costs used in the model are shown in Table 2. Sensitivity analysis was performed on all variables, using the ranges shown in Table 2.
Other assumptions of the model were that the surgeon performing the hysterectomy would have the ability to perform and interpret the results of cystoscopy, that cystoscopy is 100% sensitive and specific in detecting ureteral injuries, and that cystoscopy itself is not associated with an increase in morbidity.
The mean estimated cost of performing cystoscopy at the time of TAH was $139, and the average cost of forgoing cystoscopy was $56, for an average marginal cost (the additional cost of cystoscopy) of $83 per hysterectomy. At the base ureteral injury rate estimate of 0.5%, this translated into a marginal cost of $16,600 to prevent one unrecognized ureteral injury.
The threshold rate of ureteral injury (defined as the rate at which the marginal cost is $0) was 1.5%. At an injury rate above this threshold, performing cystoscopy was cost-saving. At the threshold rate, one would need to perform 67 cystoscopies to identify one ureteral injury intraoperatively, and the cost of performing a cystoscopy would be equal to the cost of not performing a cystoscopy. For illustrative purposes, we calculated the marginal cost or savings at two extreme rates of ureteral injury. At a low ureteral injury rate of 0.2%, one would have to perform an average of 500 cystoscopies to identify one ureteral injury intraoperatively, at a marginal cost of $108 per hysterectomy, or $54,000 per ureteral injury diagnosed intraoperatively. At a ureteral injury rate of 2%, one would have to perform 50 cystoscopies to identify one ureteral injury, at a cost savings of $44 per hysterectomy, with a cost savings of $2200 per ureteral injury.
The effect of various ureteral injury rates on the cost-effectiveness of universal cystoscopy is presented in Figure 2. Both lines are constructed by plotting the average cost per hysterectomy at various ureteral injury rates. Because one line represents the cost associated with universal cystoscopy and the other line represents the cost associated with not performing cystoscopy, the difference between the lines is the marginal cost of cystoscopy. At a very low rate of ureteral injury, the average cost per hysterectomy associated with forgoing cystoscopy was very small because the high postoperative repair costs in a very few cases were averaged over many hysterectomies. However, the cystoscopy arm of the tree had a cost of at least $125 associated with it because every hysterectomy in the cystoscopy arm incurs this fixed cost. Therefore, at low rates of injury, performing cystoscopy was more expensive than not performing cystoscopy. However, at high rates of injury, the high postoperative repair costs associated with not performing routine cystoscopy occurred in more cases, so that the ratio of cases with injury to those without was higher. Thus, the cost per hysterectomy is much higher than at a lower ureteral injury rate. In contrast, the cystoscopy arm of the tree had the fixed cost of $125 per hysterectomy with only the small cost associated with intraoperative repair. Thus, the cost per hysterectomy is relatively low. Therefore, at high rates of injury, not performing cystoscopy was more costly than performing cystoscopy, and, in fact, cystoscopy was cost-saving. Figure 2B reflects the same data in terms of cost per ureteral injury. Again, routine cystoscopy was expensive at low rates of injury, approached $0 in additional cost as the threshold injury rate of 1.5% was approached, and became cost-saving at injury rates greater than the threshold rate.
The model for TVH and LAVH was constructed to account for the additional costs associated with conversion to laparotomy if a ureteral injury requiring reimplantation was diagnosed. For this reason, the threshold ureteral injury rate is higher with this model, at 2%. Once again, for illustrative purposes, we calculated the marginal cost or savings at two extreme rates of ureteral injury (Table 3). At a ureteral injury rate of 0.2%, one would need to perform 500 cystoscopies to prevent one unidentified ureteral injury, at a cost of $113 per case. Overall, cystoscopy would cost $56,500 per ureteral injury diagnosed at this low rate of injury. At the other extreme, with a ureteral injury rate of 5%, one would need to perform 20 cystoscopies to identify one case of ureteral injury, at an average cost savings of $184 per case.
The graph of the sensitivity analysis on the ureteral injury rates with TVH and LAVH is similar to that with TAH. However, because the cystoscopy arm is more costly for the vaginal procedures, owing to the additional cost of conversion to laparotomy, the marginal cost and threshold injury rate are increased (Figure 3).
We next evaluated the other variables in the model to determine their impact on the model. The four variables with the largest effect are shown in Figure 4. The rate of ureteral injury had the greatest impact by far on the cost-effectiveness of routine cystoscopy, followed by the cost of ureteral reimplantation at the time of the original procedure. The rate of silent renal death had the third greatest impact on cost-effectiveness; however, this was a relatively minor impact even when varied from 0 to 50%. The need for percutaneous nephrostomy was fourth most important, but actually had a very small impact on the overall cost-effectiveness.
The estimate of the cost of cystoscopy also affected in a linear fashion the cost per injury prevented. For abdominal hysterectomy, at a cost of $300 per cystoscopy, the cost per injury prevented at an injury rate of 0.5% was $51,565; and at $500 per cystoscopy, the cost was $91,565 per injury prevented. However, if the injury rate was assumed to be 1.5%, then the estimated cost per injury prevented for a cystoscopy costing $300 was $11,565, and for cystoscopy costing $500, the estimated cost per injury prevented was $24,898.
Numerous surgical techniques have been suggested for the primary prevention of ureteral injury during pelvic surgery.10,13,15 Routine cystoscopy at the time of hysterectomy is an example of secondary prevention, when injury is detected at an early stage to allow early treatment. Our model for TAH suggests that, given our estimate of a ureteral injury rate of 0.5%, 200 cystoscopies would be needed to diagnose one ureteral injury intraoperatively and would cost a total of $16,600 for each injury diagnosed. However, the marginal cost-effectiveness improves as the rate of ureteral injury increases, as might be expected in difficult operative cases.
Our cost estimates were derived from average reimbursements for associated Diagnostic Related Groups in the Health Care Financing Administration Medicare Provider Analysis and Review database for 1998 and from estimates of costs from Duke University Medical Center. Because these estimates bear an uncertain relation to actual costs, sensitivity analyses were performed across wide ranges to determine the impact of these cost estimates on the model. The sensitivity analysis found that the cost-effectiveness of routine intraoperative cystoscopy depends most on the rate of ureteral injury. The cost of reimplantation at a separate admission and the rate of silent renal death also influence cost-effectiveness, but to a much lesser degree. The cost of cystoscopy also plays a role at very high estimates of cost. Our review of the literature highlighted the fact that most ureteral injuries are unsuspected and many, if not most, occur during routine surgical procedures, limiting the surgeon's ability to make a decision based solely on his or her estimated ureteral injury risk in a given case.
This study was limited because it is based on a mathematical model with uncertainty surrounding some variables. However, we attempted to identify the effects of the uncertainty on our estimates by sensitivity analysis. Cost-effectiveness depends on perspective. The available data led us to choose a short-term, hospital-based perspective, and we therefore did not incorporate variables such as patient pain and suffering, time lost from work, physician malpractice, and physician anguish associated with litigation.
We chose not to include an estimate of the cost of litigation and negligence settlements resulting from intraoperative ureteral injury because of a lack of data and methodologic uncertainty surrounding how best to include malpractice costs in health-economic analyses. However, if one assumes that malpractice costs increase the overall costs of ureteral injury, including these costs would only have lowered the threshold ureteral injury rates above which cystoscopy would, in fact, be cost-saving. Potentially limiting exposure to medicolegal liability by intraoperative recognition of a ureteral injury is certainly worthy of consideration. One recent review of malpractice cases secondary to ureteral injury cited only cases that were not diagnosed during the original operation.35
The threshold rates that we identified are the ureteral injury rates at which the cost of universal cystoscopy is equal to the cost of no cystoscopy. They are not thresholds below which cystoscopy is not “cost-effective”; indeed, most accepted preventive interventions, including Papanicolaou tests, mammography, colon cancer screening, and cholesterol testing, have net positive costs. Performing cystoscopy is similar to any other procedure that is associated with often competing benefits, risks, and costs. Pelvic surgeons must decide for themselves which procedures warrant a diagnostic intraoperative cystoscopy.
Our assumptions about the effectiveness and costs of cystoscopy are dependent on competence of pelvic surgeons with the technique of cystoscopy and ready availability of cystoscopic equipment. Obviously, needing to call in a colleague to perform the surgery or causing injury from the procedure itself would rapidly increase the costs. One benefit of universal cystoscopy is that, as with all procedures, repetition would enhance skills and outcomes. Hospitals in which gynecologists do not have cystoscopic privileges, or where such equipment is not readily available to gynecologic surgeons, would also be unlikely to see benefit in our base-case estimates. One potential use of this model would be as a tool for hospitals to examine their policies regarding cystoscopy, especially in the setting of detection of ureteral injuries, using their own estimates of injury rates and costs. Another limitation of our model was that we assumed no morbidity from the cystoscopy. Although any invasive surgical procedure has associated risks (eg, infection, bladder or urethral injury), diagnostic cystoscopy is relatively safe in trained hands. Again, this increase in morbidity would be associated with an increased overall cost of cystoscopy, which would increase the overall costs of prophylactic cystoscopy.
This study examined the cost-effectiveness of universal cystoscopy only during hysterectomy, not during other procedures such as reconstructive pelvic operations. Because the overall rate of ureteral injury had the greatest impact on cost-effectiveness, it seems reasonable to recommend universal cystoscopy to confirm bilateral ureteral patency for all gynecologic procedures with associated high rates of ureteral injuries, such as pelvic reconstructive procedures.36 Shull et al37 reported a 1.6% ureteral injury rate during uterosacral ligament suspension for vaginal vault prolapse. In a recent presentation, Barber et al reported an injury rate of 4% with the same procedure (American Urogynecologic Society Meeting, 1999). During such surgical procedures that place the ureter at higher risk, cystoscopy is not only helpful in preventing renal death and ureterovaginal fistula formation, but it may actually save money.
We conclude that routine cystoscopy at the time of hysterectomy may be cost-saving when the rate of ureteral injury is above 1.5% for TAH and above 2% for TVH and LAVH. At rates lower than these threshold values, cystoscopy has a net positive cost. Whether this cost is worthwhile is a subject for further research and discussion among surgeons, hospitals, professional organizations, payers, and patients.
1. Haug JN, Politser PE. Socio-economic factbook for surgery. Chicago: American College of Surgeons, 1987.
2. Goodno JA Jr, Powers TW, Harris VD. Ureteral injury in gynecologic surgery: A ten-year review in a community hospital. Am J Obstet Gynecol 1995;172:1817–22.
3. Harkki-Siren P, Sjoberg J, Tiitinen A. Urinary tract injuries after hysterectomy. Obstet Gynecol 1998;92:113–8.
4. Thompson JD, Benigno BB. Vaginal repair of ureteral injuries. Am J Obstet Gynecol 1971;111:601–10.
5. Solomons E, Levin EJ, Bauman J, Baron J. A pyelographic study of ureteric injuries sustained during hysterectomy for benign conditions. Surg Gynecol Obstet 1960;111:41–48.
6. Stanhope CR, Wilson TO, Utz WJ, Smith LH, O'Brien PC. Suture entrapment and secondary ureteral obstruction. Am J Obstet Gynecol 1991;164:1513–9.
7. Saidi MH, Sadler RK, Vancaillie TG, Akright BD, Farhart SA, White AJ. Diagnosis and management of serious urinary complications after major operative laparoscopy. Obstet Gynecol 1996;87:272–6.
8. Tamussino KF, Lang PF, Breinl E. Ureteral complications with operative gynecologic laparoscopy. Am J Obstet Gynecol 1998;178:967–70.
9. Ribeiro S, Reich H, Rosenberg J, Guglielminetti E, Vidali A. The value of intra-operative cystoscopy at the time of laparoscopic hysterectomy. Hum Reprod 1999;14:1727–9.
10. Dowling RA, Corriere JN Jr, Sandler CM. Iatrogenic ureteral injury. J Urol 1986;135:912–5.
11. Fry DE, Milholen L, Harbrecht PJ. Iatrogenic ureteral injury. Options in management. Arch Surg 1983;118:454–7.
12. Witters S, Cornelissen M, Vereecken R. Iatrogenic ureteral injury: Aggressive or conservative treatment. Am J Obstet Gynecol 1986; 155:582–4.
13. Spence HM, Boone T. Surgical injuries to ureter. JAMA 1960;176:1070–5.
14. Higgins CC. Ureteral injuries during surgery. A review of 87 cases. JAMA 1967;199:82–8.
15. Conger K, Beecham CT, Horrax TM. Ureteral injury in pelvic surgery. Obstet Gynecol 1954;3:343–57.
16. St. Martin EC. Ureteral injuries in gynecologic surgery. J Urol 1953;70:51–7.
17. Mattingly RF, Borkowf HI. Acute operative injury to the lower urinary tract. Clin Obstet Gynecol 1978;5:123–49.
18. Blandy JP, Badenoch DF, Fowler CG, Jenkins BJ, Thomas NW. Early repair of iatrogenic injury to the ureter or bladder after gynecological surgery. J Urol 1991;146:761–5.
19. Hoch WH, Kursh ED, Persky L. Early, aggressive management of intraoperative ureteral injuries. J Urol 1975;114:530–2.
20. Flynn JT, Tiptaft RC, Woodhouse CR, Paris AM, Blandy JP. The early and aggressive repair of iatrogenic ureteric injuries. Br J Urol 1979;51:454–7.
21. Badenoch DF, Tiptaft RC, Thakar DR, Fowler CG, Blandy JP. Early repair of accidental injury to the ureter or bladder following gynaecological surgery. Br J Urol 1987;59:516–8.
22. Kihl B, Nilson AE, Pettersson S. Ureteroneocystostomy in the treatment of postoperative uretero-vaginal fistula. Acta Obstet Gynecol Scand 1982;61:341–6.
23. Hedlund H, Lindstedt E. Urovaginal fistulas: 20 years of experience with 45 cases. J Urol 1987;137:926–8.
24. Wiskind AK, Thompson JD. Should cystoscopy be performed at every gynecologic operation to diagnose unsuspected ureteral injury? J Pelv Surg 1995;1:134–7.
25. Simel DL, Matchar DB, Piscitelli JT. Routine intravenous pyelograms before hysterectomy in cases of benign disease: Possibly effective, definitely expensive. Am J Obstet Gynecol 1988;159:1049–53.
26. Dicker RC, Greenspan JR, Strauss LT, Cowart MR, Scally MJ, Peterson HB, et al. Complications of abdominal and vaginal hysterectomy among women of reproductive age in the United States. The Collaborative Review of Sterilization. Am J Obstet Gynecol 1982;144:841–8.
27. Amirikia H, Evans TN. Ten-year review of hysterectomies: Trends, indications, and risks. Am J Obstet Gynecol 1979;134:431–7.
28. Falk HC, Bunkin IA. Ureteral injuries. Obstet Gynecol 1954;4:4–20.
29. Holloway HJ. Injury to the urinary tract as a complication of gynecologic surgery. Am J Obstet Gynecol 1950;60:30–40.
30. Newell QU. Injury to ureters during pelvic operations. Ann Surg 1939;109:981–6.
31. Neuman M, Eidelman A, Langer R, Golan A, Bukovsky I, Caspi E. Iatrogenic injuries to the ureter during gynecologic and obstetric operations. Surg Gynecol Obstet 1991;173:268–72.
32. Boyd ME, Groome PA. The morbidity of abdominal hysterectomy. Can J Surg 1993;36:155–9.
33. Copenhauer EH. An analysis of indications and complications among 1000 operations. Am J Obstet Gynecol 1962;84:123–8.
34. Edwards EA, Beebe RA. Vaginal hysterectomy. Surg Gynecol Obstet 1949;89:91–9.
35. Brudenell M. Medico-legal aspects of ureteric damage during abdominal hysterectomy. Br J Obstet Gynaecol 1996;103:1180–3.
36. Harris RL, Cundiff GW, Theofrastous JP, Yoon H, Bump RC, Addison WA. The value of intraoperative cystoscopy in urogynecologic and reconstructive pelvic surgery. Am J Obstet Gynecol 1997;177:1367–71.
37. Shull BL, Benn SJ, Kuehl TJ. Surgical management of prolapse of the anterior vaginal segment: An analysis of support defects, operative morbidity, and anatomic outcome. Am J Obstet Gynecol 1994;171:1429–39.