The risk of vesicovaginal and urethrovaginal fistula after hysterectomy performed in the English National Health Service—a retrospective cohort study examining patterns of care between 2000 and 2008


  • P Hilton,

    1.  Newcastle upon Tyne Hospitals NHS Foundation Trust, Newcastle, UK
    Search for more papers by this author
  • DA Cromwell

    1.  Office of Research and Clinical Audit, Royal College of Obstetricians and Gynaecologists, London, UK
    2.  Department of Health Services Research and Policy, London School of Hygiene and Tropical Medicine, London, UK
    Search for more papers by this author

P Hilton, Consultant Gynaecologist & Urogynaecologist, Directorate of Women’s Services, Level 5, Leazes Wing, Royal Victoria Infirmary, Newcastle upon Tyne, NE1 4LP, UK. Email


Please cite this paper as: Hilton P, Cromwell D. The risk of vesicovaginal and urethrovaginal fistula after hysterectomy performed in the English National Health Service—a retrospective cohort study examining patterns of care between 2000 and 2008. BJOG 2012;119:1447–1454.

Objective  To estimate rates of vesicovaginal and urethrovaginal fistula among women undergoing hysterectomy by indication and type of procedure, and to assess trends in risk over time.

Design  Retrospective cohort using data from Hospital Episode Statistics.

Setting  English National Health Service (NHS) hospitals.

Population  Women undergoing hysterectomy for selected common conditions in English NHS hospitals between January 2000 and December 2008.

Methods  Unadjusted rates of urogenital fistula were calculated by type of procedure and indication. Logistic regression was used to assess whether the risk of fistula was associated with age, or had changed over time.

Main outcome measure  Rate of urogenital fistula (vesicovaginal and urethrovaginal fistula) within 1 year of hysterectomy.

Results  Among 343 771 women undergoing hysterectomy, the overall rate of fistula was 1 in 788. The rate varied by indication and procedure, being highest following radical hysterectomy for cervical cancer (1 in 87; 95% CI 61–128) and lowest following vaginal hysterectomy for prolapse (1 in 3861; 95% CI 2550–6161). After total abdominal hysterectomy for endometriosis, menstrual problems or fibroids, the risk of fistula was lower in women aged 50 years or over than in women under 40 years (adjusted odds ratio 0.61; 95% CI 0.38–0.98). The overall rate of fistula increased by 46% during the study period.

Conclusions  The risk of urogenital fistula was associated with type of hysterectomy and indication; the risk increased during the study period, and was lower after hysterectomy for benign conditions in women aged 50 years or over.


The prevention and management of vesicovaginal fistula remains a major global public health challenge.1 It has been estimated that there are currently 2 000 000–3 000 000 women with untreated fistula worldwide, although over 95% of these are of obstetric aetiology and are in the developing world.2 In the developed world, urogenital fistulae are relatively uncommon, and publically available Hospital Episode Statistics (HES) for England suggest that approximately 105 cases have been repaired surgically each year over the last decade.3

Chassar Moir4 reviewed the position of vesicovaginal fistula in Britain in 1973, and Lee et al.5 did the same in the USA in 1988. The largest recent UK case series are those reflecting tertiary-care practices, published by Ockrim et al.6 in 2009 (41 cases), and Hilton7 in 2011 (348 cases). In the latter series, two-thirds of cases were of surgical aetiology, and the remainder were evenly distributed between obstetric, radiation and miscellaneous or traumatic causes. Overall, almost half of this series were associated with hysterectomy.7

The number of hysterectomy procedures undertaken in the UK has been falling over the last decade, with the increasing availability and acceptability of nonsurgical treatments for menorrhagia in particular.8 However, with the number of surgically treated urogenital fistulae remaining fairly constant over the same period, this suggests that the risk of urogenital fistula following hysterectomy may be increasing.7 In this study, we describe rates of vesicovaginal fistula and urethrovaginal fistula within 1 year of hysterectomy among women treated in English NHS Trusts between January 2000 and December 2008, and examine whether the risk of a urogenital fistula is affected by the indication, the type of procedure carried out, and the age at surgery. We also investigate whether the risk of fistula has changed over time.


The study used data from the HES database, which contains records of all patient admissions to NHS hospitals in England. Its records capture patient demographics, hospital administrative information, diagnostic information using the International Classification of Diseases (ICD), 10th revision,9 and operative procedures using the UK Office for Population Censuses and Surveys classification (OPCS), 4th revision.10

Patient selection

This study included adult women (18 years and over) who underwent hysterectomy for selected common indications in English NHS hospitals between 1 January 2000 and 31 December 2008. The study was restricted to elective procedures and distinguished between four types of hysterectomy: total abdominal hysterectomy (TAH) (OPCS code: Q07.4), subtotal abdominal (STH) (Q07.5), vaginal (VH) (Q08.9), and radical abdominal hysterectomy (RAH) (Q07.1-07.3). The selected benign conditions were endometriosis (ICD10 code: N80), uterine fibroids (D25), genital prolapse (N81) and menstrual disorders (N92–N94). Four ‘malignant’ conditions were also selected, namely, cervical carcinoma in situ (D06), cervical cancer (C53), uterine cancer (C54, C55) and ovarian cancer (C56). The analysis was restricted to women under 55 years of age if their primary diagnosis was endometriosis or menstrual irregularities. Further, the analysis also excluded radical hysterectomies coded to women with benign diagnoses, and combinations of conditions and procedures for which there were fewer than 500 cases.

Women were defined as having a postoperative fistula if the ICD10 codes N82.0 (vesicovaginal fistula) or N82.1 (other urinary-genital tract fistula) were found in the diagnostic fields of the index procedure episode or any subsequent episode that occurred within 1 year of the index procedure. The majority of fistulae occurred within 6 months of the index procedure. Surgical repair of fistulae was examined using the most pertinent OPCS codes, namely, repair of vesicovaginal fistula (OPCS P25.1), repair of urethrovaginal fistula (P25.2) and construction of an ileal conduit (M19.1).

Statistical analysis

The unadjusted incidence rate of urogenital fistula was expressed as the percentage of fistulae within 1 year of hysterectomy; rates were stratified by type of procedure and indication. We estimated the 95% confidence intervals (CI) assuming that the number of events followed a Poisson distribution. Fisher’s exact test was used to judge the statistical significance of differences between unadjusted rates.

We also examined whether the risk of urogenital fistula had changed over time and was related to patient age at the time of procedure. The primary analysis was restricted to TAH procedures for endometriosis, fibroids or menstrual conditions because of the small number of fistulae recorded for other procedure–indication combinations. Two additional analyses were performed. One extended the first analysis by including both VH and TAH procedures for these three benign indications. The other examined the associations among women undergoing TAH for a malignant indication. Logistic regression was used to calculate adjusted odds ratios (OR) and 95% CI. All P values were two-sided, and those <0.05 were judged to be statistically significant. All analyses were performed using STATA version 11 (StataCorp LP, College Station, TX, USA). The formulation of this report has followed the STROBE Statement.11


Between 1 January 2000 and 31 December 2008, there were 351 505 hysterectomies performed in English NHS hospitals for the selected indications (Figure 1). Of the 292 250 operations performed for benign conditions, 3262 (1.1%) were excluded because the procedure was coded as a radical hysterectomy (abdominal or vaginal) and 290 (0.1%) were excluded because of the small number criterion. A further 2645 (2.5%) of the 106 440 women with endometriosis or menstrual conditions were excluded for being 55 years or older. Of the 59 255 procedures performed for malignant conditions, 1537 (2.6%) were excluded because of the small number criterion. The characteristics of the remaining 343 771 women are summarised in Table 1.

Figure 1.

 Flow chart of included and excluded patients and procedures.

Table 1.   Patient characteristics of women undergoing hysterectomy in NHS hospitals for selected conditions between 2000 and 2008 (after exclusions as described in text)
N92 to N94
Benign conditions
All patients87 32924 94778 84894 929
Average age
Age group
 Under 405%45%45%14%
 70 and over23%2%
Type of hysterectomy
 Abdominal subtotal2331663411 546
 Abdominal total238619 21653 85173 774
 Vaginal84 943340018 3639609
  Cervical cancer in situ
Cervical cancer
Uterine cancer
C54, C55
Ovarian cancer
Malignant conditions
All patients3886532734 26114 244
Average age
Age group
 Under 4038%44%1%6%
 70 and over3%6%34%24%
Type of hysterectomy
 Abdominal subtotal1080
 Abdominal total2890262731 03213 164
 Abdominal radical2700538

Number of hysterectomy procedures by indication

The number of hysterectomies performed for the different indications varied over the 9-year period (Figure 2). For women with menstrual conditions, endometriosis, or fibroids, the number of procedures fell by 37%, 30% and 15%, respectively. The number of procedures for cervical cancer and carcinoma in situ fell by 13% and 39%. In contrast, the number of hysterectomies for uterine cancer, ovarian cancer and prolapse rose by 37%, 18% and 11%, respectively. Overall, the annual number of hysterectomies declined by 16% from 43 014 in 2000 to 36 189 in 2005, before increasing by 5% to 37 923 in 2008. Between 2000 and 2008, the proportion of TAH procedures decreased from 60% to 53%, whereas VH increased from 34% to 39%. The relative frequency of STH and RAH remained comparatively static.

Figure 2.

 Number of hysterectomy procedures in English NHS hospitals for selected conditions between 2000 and 2008.

Rates of urogenital fistula by procedure and condition

There were 436 women who had a diagnosis of vesicovaginal fistula or urethrovaginal fistula recorded within a year of their hysterectomy, giving an overall rate of 1 in 788 (95% CI 718–868). However, the risk posed to women differed across the various combinations of procedure and indication (Table 2). For women undergoing VH for prolapse, the estimated rate was 1 in 3861 (95% CI 2550–6161). Among women undergoing a TAH for the other three benign conditions, the overall rate was 1 in 540 (272/146 841; 95% CI 479–540), whereas for women who underwent STH, the rate was 1 in 2279 (9/20 511; 95% CI 1205–5000). For both these abdominal procedures, the differences in the fistula rates across the three benign conditions were not statistically significant (Fisher’s Exact; = 0.9 for TAH and = 0.17 for STH). In contrast, the absolute risk for women undergoing VH differed slightly across the three conditions (Fisher’s Exact, = 0.04), the rate ranging from 1 in 486 for endometriosis to 1 in 1922 for fibroids.

Table 2.   Rates of fistula within 1-year of undergoing hysterectomy, by condition and type of procedure
ConditionsType of hysterectomyNo. of proceduresNo. of FistulaeFistula rate (%)95% confidence interval (%)
  1. *One-sided confidence interval.

Benign conditions
ProlapseAbdominal total238610.040.000.23
Vaginal84 943220.030.020.04
EndometriosisAbdominal subtotal233100.000.000.16*
Abdominal total19 216360.190.130.26
MenstrualAbdominal subtotal663410.020.000.08
Abdominal total53 8511030.190.160.24
Vaginal18 363230.130.080.19
FibroidsAbdominal subtotal11 54680.070.030.14
Abdominal total73 7741330.180.150.21
‘Malignant’ condition
Cervical carcinoma in situAbdominal total289050.170.060.40
Cervical cancerAbdominal total2627210.800.491.22
Abdominal radical2700311.150.781.63
Uterine cancerAbdominal total31 032280.090.060.13
Abdominal radical53830.560.111.63
Ovarian cancerAbdominal subtotal108010.090.000.52
Abdominal total13 16470.050.020.11

The risk of vesicovaginal fistula or urethrovaginal fistula was greatest among those women who had cervical cancer, with TAH and RAH both having estimated rates of around 1 in 100 (Table 2). Among women undergoing TAH, the rate was roughly five times greater than the rates for women with cervical carcinoma in situ (relative risk 4.6; 95% CI 1.7–12.2; P < 0.001). Among the other uterine cancers, RAH also posed an increased risk of urogenital fistula, but overall the rates of fistula were of the same order of magnitude as the rates for the benign conditions.

Effect of age and year of surgery

For the benign conditions of endometriosis, fibroids and menstrual disorders, there were noticeable changes in the number of fistulae after TAH across the age categories and by year of surgery (Table 3). In terms of age, the rate of fistula was stable until women passed 50 years of age, after which it fell by around 40%. The rate of fistula increased from 0.15% in 2000–02 to 0.22% in 2006–08, a rise of 46%. Adjustment for possible confounding by indication and time period did not alter these patterns. Extending the model to include both VH and TAH for these benign conditions produced comparable results for the estimated relationship between rate of fistula and age categories and year of surgery. For patients with malignant conditions, there was no association between the adjusted OR for fistula by year of surgery (= 0.97) nor by age category (= 0.86).

Table 3.   Estimated association between rates of fistula and various risk factors for women undergoing total abdominal hysterectomy for fibroids, endometriosis and menstrual disorders
 No. of proceduresNo. of fistulaeRate (%)Adjusted odds ratio95% confidence interval P value
Fibroids73 7741330.181  0.99
Endometriosis19 216360.190.980.671.43
Menstrual disorders53 8511030.191.010.761.32
Age group (years)
Under 4042 873770.181  0.037
40–4443 957920.211.140.831.55
45–4937 312780.211.140.821.58
50 plus22 699250.110.610.380.98
Year of surgery
2000–0256 561830.151  0.033
2003–0547 535970.201.381.031.85
2006–0842 745920.221.441.071.94

Surgical repair

The majority of women with fistula underwent one of the repair operations. Urogenital fistula repair procedures (P25.1, P25.2) were found for 71% of fistulae among women with benign conditions (241/339). None of these women had a urinary diversion. For women with malignant conditions, fistula repair was identified for 41% (41/97) of women. Another three women (all of whom had RAH for cervical cancer) had a urinary diversion.


This study of elective hysterectomies in English NHS hospitals has established the overall rate of lower urinary–genital tract fistulae within 1 year of surgery, of 0.13% or 1 in 788. The rate was highest among women who had either an RAH or TAH for cervical cancer, with approximately 1 in 100 experiencing a fistula. The most common type of hysterectomy undertaken for fibroids, endometriosis and menstrual disorders was TAH, and resulted in an estimated fistula rate of 1 in 540. For the same diagnostic groups, the rate was 1 in 2279 for STH. The lowest rate of fistula following hysterectomy was observed for VH for prolapse (1 in 3861).

The Cochrane review of approaches to hysterectomy for benign disease found no difference in the rate of fistula following abdominal, vaginal and laparoscopic routes;12 and the VALUE study, a non-randomised cohort study of hysterectomy from the UK, although not reporting on the development of fistula, found no difference in the rate of operative bladder injury between abdominal, vaginal and laparoscopic routes.13 The national cohort study from Finland on the other hand reported a higher rate of both bladder injury (relative risk 11.4) and subsequent fistula formation (relative risk 2.2) following laparoscopic procedures.14 Unfortunately, the OPCS-4 coding used in HES does not allow easy separation between open and laparoscopic hysterectomies. There is an option to use an additional code for ‘laparoscopic approach’ alongside the codes for hysterectomy, although an evaluation of OPCS coding completeness in other situations has shown that laparoscopic procedures are under-reported.15 Given the range of surgical techniques and coding systems used during our study period, we do not feel that much of use can be stated about the risks of fistulae associated with laparoscopic hysterectomy procedures.

It is often stated that distorted anatomy (e.g. from fibroids or malignancy), abnormal tissue adhesion (e.g. from previous caesarean section or endometriosis) or previous radiation therapy increase the risk of urinary tract injury and fistula formation following hysterectomy.16 The HES database does not include radiotherapy treatments, and given the average age of these women (51 years) does not have sufficient historical information to capture obstetric history. We did not find the rate of fistula to be increased in women undergoing abdominal hysterectomies for fibroids or endometriosis compared with women with menstrual problems (although there was variation in risk for women having vaginal procedures for these benign indications). The HES database gives no information on the stage of endometriosis or the size of fibroids, making it difficult to comment on their significance as risk factors.

Calculating precise rates of post-hysterectomy fistula requires a large sample size and a complete data set. Both of these criteria were met in this study, with the data extracted from the HES database containing 343 771 procedures in total, and over 10 000 patients for many of the indication–procedure combinations analysed. However, in any study using routine data, there are potential limitations arising from coding omissions and inaccuracy. The quality of the clinical coding in HES has been shown to vary between NHS trusts,17 although the overall quality is considered to be high in relation to operations and procedures.18

Coding accuracy could have influenced the identification of hysterectomy procedures, and the classification by type. However, studies that have compared medical notes to routine data for hysterectomy have found very good agreement.19 In addition, to remove records affected by coding errors, we restricted our analysis to patient records with (a) expected combinations of diagnosis and procedure, and (b) age ranges consistent with the diagnosis.

Coding accuracy and completeness could have also affected our estimated incidence of fistula. In series of fistulae from the developed world, between 50% and 90% follow hysterectomy,7,20 and postoperative fistula rates between 0.1% and 4.0% are reported.16,21 The overall rate described in our study (1 in 788) lies within this range of estimates, and is consistent with other large-scale national studies.14,21 There is no reason to suspect that the coding of fistula would be more complete for some procedures than for others, so the estimated relative rates among the different patient groups are likely to be robust.

Comparison with other studies

Our study is the first to report a significantly lower risk of urogenital fistula for women over 50 years of age. This is at odds with a study by Forsgren et al.,21 who reported the rate of ‘fistula disease’ to be higher in women over 50 years of age. However, their study did not separate intestinal from urinary fistulae, and almost 60% did not involve the urinary tract. The same group recently examined the extent of association between ‘fistula disease’ in women with diverticulitis undergoing hysterectomy and reported the main associations being with intestino-genital and uro-intestinal fistulae.22 It is conceivable therefore that the increase in fistulae reported in older women in these studies relates more to bowel than urinary fistulae, and to the prevalence of inflammatory bowel disease than to hysterectomy. The VALUE study found a reduction in overall serious morbidity with increasing age at operation, although this was limited to women with fibroids and to women undergoing vaginal hysterectomy.13

Our analysis provides no insight into the reasons for lower risk of urogenital fistula for women over 50 years of age. We can speculate that the reduced vascularity and uterine size in the estrogen-deficient postmenopausal pelvis might reduce the risk of operative injury or postoperative haematoma formation. The finding is unlikely to reflect confounding because of the exclusion of women over 55 years old coded as having hysterectomy for endometriosis and menstrual problems from the analysis because indication was an explanatory factor in the multiple logistic regression. Moreover, a lower risk of urogenital fistula for women over 50 years of age was still apparent if the analysis was restricted to women with uterine fibroids whose age range was not restricted. A quarter of the women were aged over 50 years.

The national cohort studies from Sweden,21 and Finland,14 both showed lower overall rates of fistula following hysterectomy for benign indications, most notably for TAH (1 in 1019 and 1 in 959, respectively, compared with one in 540 in this series). One possible explanation for the difference is that the Swedish study covered the period 1973–2003 and the Finnish study 1990–95 whereas our study covered a later period (2000–08). We also observed a 46% increase in risk of fistula following hysterectomy for benign conditions over the study period. One explanation for this could be an increase in the proportion of more difficult procedures among all hysterectomies. Between 2000 and 2008, the total number of hysterectomy procedures fell by 12% and those for the main benign indications (excluding prolapse) fell by 27%. This was mainly a result of the fall in the number of hysterectomies for menorrhagia, a trend that has been ongoing since the mid-1990s.23

Another potential contributory factor is the changes in surgical training and practice in gynaecology. A consequence of the reduced number of hysterectomy procedures is that trainees, in the UK at least, obtain less surgical experience during their training. It was calculated in 2001 that even if every hysterectomy carried out in the NHS were carried out by a trainee (and the majority are not), each would undertake only 27 procedures per year during their training.24 The reduction of the length of the working week resulting from the ‘New Deal’ and the European Working Time Directive,25,26 and reduced years in training implemented following the Report of the Working Group on Specialist Medical Training and ‘Modernising Medical Careers’,27,28 both add to this problem. Trainees themselves perceive this to be an issue. Although the quality of supervision of surgical training is reported to be of a good standard, trainees’ satisfaction with operative teaching declined consistently between 1995 and 2008, and operating volume was seen as a significant factor in this.29


This is the largest reported cohort study on the risk of urogenital fistula following hysterectomy. Overall, 1 in 788 hysterectomies in the NHS in England is followed by vesicovaginal fistula or urethrovaginal fistula. The rate is highest following RAH for cervical cancer (1 in 87), and lowest following VH for prolapse (1 in 3861). The most common procedure–indication combination is TAH for benign indications, which carries a risk of fistula formation of 1 in 540. Whereas urogenital fistula remains an uncommon complication of hysterectomy,30 our study also identified a trend of increasing surgical morbidity within the NHS, of which all concerned in the provision of service and training should be aware.

Disclosure of interests

PH has no financial interests to declare. He is a faculty member on ICUD-SIU 1st International Consultation on Obstetric Vesicovaginal Fistula—complications committee (2010/11) and on the ICUD-EUA 5th International Consultation on Incontinence—fistula committee (2011/12). DC has no interests to declare.

Contribution to authorship

PH conceived the study and contributed to its design, contributed to analysis and writing of the paper, and approved the final version for publication. DC contributed to study design, data extraction and analysis, writing of the paper, and approved the final version for publication.

Details of ethics approval

The study is exempt from UK NREC approval because it involved analysis of an existing data set of anonymised data for service evaluation. Approvals for the use of HES data were obtained as part of the standard Hospitals Episode Statistics approval process.




We thank the Department of Health for providing the Hospital Episode Statistics data used in this study.

Journal club

Discussion points

  • 1 Background: Describe the risks of hysterectomy.1 Discuss the implications of vesicovaginal and urethrovaginal fistula following hysterectomy and their clinical presentation in practice.
  • 2 Methods: Discuss the possible benefits and pitfalls of using the Hospital Episode Statistics (HES) database as a source to investigate outcomes of surgery in the NHS. Compare HES with central and local clinical databases, with particular reference to provision of demographic data, case ascertainment, completeness of data, availability of clinical detail (endometriosis stage, clinical history). 2,3 Discuss coding issues including provision or not of codes for recent and novel procedures (e.g. laparoscopic hysterectomy). (Optional advanced task): Regression analysis was conducted at patient-level data and not hospital-level aggregated data. Compare the two approaches. Which one would account for variation in rates per individual surgeon? Can the two approaches be combined?
  • 3 Results and implications: The authors describe an increase in the overall rate of fistula of 46% during the study period. A possible contributory factor to this increase proposed by the authors is the change in surgical training and practice in gynaecology and the reduced number of hysterectomies performed by trainees—discuss.Would the findings of this paper affect your counselling of women for hysterectomy depending on their indication for surgery? (Data S1)

Disclosure of interest

Rebecca Simms has been conducting research on maternity databases.

D Siassakos & R Simms

RiSQ, School of Clinical Sciences, University of Bristol, UK