Vaginal Complications after Cystectomy: Results from a Medicare Sample

Abstract

Purpose:

Cystectomy with a vaginal sparing approach may be associated with unique complications specific to the female population. The objective of this study was to estimate the incidence of vaginal complications (defined to include vaginal prolapse, vaginal fistula, dyspareunia, and vaginal cuff dehiscence/evisceration) after cystectomy and to determine risk factors for these complications.

Materials and Methods:

Women 65 or older undergoing cystectomy for any indication were identified by procedural codes in the Medicare Limited Data Set 5% sample from January 1^st 2011 to December 31^st 2017. Patients experiencing a vaginal complication after cystectomy were compared to those who did not. Demographic and biological factors that could increase likelihood of complications were identified and time to development of complications determined. Cumulative incidence was calculated using cumulative incidence function. Multivariable cause-specific Cox proportional hazards model assessed risk factors for vaginal complications.

Results:

481 women undergoing cystectomy were identified during the study period, 37.2% were less than 70 years old. The majority 378 (79%) had bladder cancer and 401 (83.4%) underwent an incontinent conduit or catheterizable channel diversion. Within two years of cystectomy, 93 patients (19.5%) had one or more complications on record. Vaginal cuff dehiscence had the highest cumulative incidence, occurring in 49 patients (10.2%). Over the entire study period (2011 – 2017), 102 women (21.2%) were diagnosed with a vaginal complication, and 27 (5.6%) received an intervention.

Conclusions:

Among women who undergo cystectomy, vaginal complications occur at rates higher than expected with over 20% of women experiencing a complication and over a quarter of those diagnosed undergoing intervention.

Introduction:

Due to advances in multimodal treatment for pelvic malignancies, life expectancy after cystectomy is increasing¹. As patients live longer, there is increasing focus on issues related to cancer survivorship and quality of life.

For women with bladder cancer, cystectomy with a vaginal sparing approach may be associated with unique complications specific to the female population: pelvic organ prolapse, vaginal dehiscence or evisceration, vaginal fistula, and dyspareunia. While several case series of these individual complications have been reported, the actual incidence remains unknown. Furthermore, vaginal complications are not typically considered as a potential long-term risk of cystectomy, however they may occur at rates higher than expected in this population.

Our primary objective is to estimate the incidence of vaginal complications (defined to include vaginal prolapse, vaginal fistula, dyspareunia, and vaginal cuff dehiscence/evisceration) after cystectomy and to determine risk factors for development of these complications. Our secondary objective is to describe the surgical approaches used to manage these complications.

Materials and Methods:

In this retrospective cohort study, women undergoing cystectomy for the following indications were identified from the Medicare 5% Limited Data Set between January 1^st, 2011-December 31^st, 2017 (Cancer of urinary tract: bladder, urethra; Cancer of reproductive organs: cervix, uterus, ovary, vagina or vulva; Cancer of colon or rectum; Benign conditions: interstitial cystitis, neurogenic bladder). Patients who experienced a vaginal complication (defined to include pelvic organ prolapse, vaginal fistula, vaginal cuff dehiscence/evisceration, or dyspareunia) after cystectomy were identified and compared to those who did not have a record of vaginal complication. This study was deemed exempt by our Institutional Review Boards. Per the Centers of Medicare & Medicaid cell suppression policy, the exact number of any sub-group or table cell with <11 patients was masked.

Population:

Current Procedural Terminology (CPT) codes for either open or minimally invasive cystectomy (51570, 51575, 51580, 51585, 51590, 51595, 51596, 51597) and the International Classification of Diseases (ICD)-9 codes (57.7, 57.71, 57.79, 68.8) were used to determine the associated procedures. (The 51999 CPT code, when billed with an ICD diagnosis code for bladder cancer, was utilized to identify robotic cystectomy cases.) Women who underwent cystectomy between January 1^st, 2011-December 31^st, 2017 were included. Men and individuals with missing gender were excluded. Women without continuous Medicare fee-for-service enrollment one year before the date of index surgery were also excluded.

Outcome:

Our primary outcome was the incidence of vaginal complications (composite outcome of any/all complications) after cystectomy. These complications included pelvic organ prolapse, vaginal fistula, vaginal cuff dehiscence/ evisceration, or dyspareunia. Vaginal fistula was defined broadly to include neobladder vaginal, peritoneal vaginal, or enterovaginal fistula. ICD-9 and ICD-10 codes were used to identify patients diagnosed with vaginal complications at any time point after cystectomy until death, loss of enrollment, or study end date (i.e. December 31^st, 2017) (Supplementary Table 1). Subsequently, CPT codes for relevant surgical and non-surgical management of pelvic organ prolapse, vaginal fistula, and vaginal dehiscence or evisceration were used to identify patients undergoing these interventions (Supplementary Table 2).

Covariates:

We identified demographic (age, race, smoking status) and biologic (indication for cystectomy, type of urinary diversion, radiation history, obesity, Charlson co-morbidity score) factors that were hypothesized to increase the likelihood of developing vaginal complications after cystectomy. Management of women with vaginal complications was characterized, and the timing of intervention relative to cystectomy was described.

Statistical Analysis:

Patient baseline characteristics were described by indication for cystectomy during the study period. Continuous covariates were presented as median [25^th, 75^th] and compared using Kruskal-Wallis test. Categorical covariates were presented as frequencies (%) and compared using Pearson’s Chi-square test.

The number of interventions for vaginal complications was reported for both the 2-year period following cystectomy as well as throughout the study period. Cumulative incidence function was used to estimate the incidence of vaginal complications over a 2-year period following cystectomy. For patients who did not experience vaginal complications and were lost to follow-up before the study end date due to loss of fee-for-service insurance coverage or death, censoring was determined to occur on the date of disenrollment or date of death. Death was determined from death date variable, while loss of follow-up due to disenrollment was determined from enrollment dates.

The effect of covariates on time to any vaginal complications was determined using cause-specific Cox proportional hazards. Univariate and multivariable Cox proportional hazards model were used to provide assessments of the independent associations of demographic and cystectomy characteristics with incident diagnosis of vaginal complications. Associations between covariates and outcome were further assessed while explicitly accounting for death as a competing event using a multivariable Fine and Gray sub-distribution hazards model. Treatments were summarized into categories: non-surgical treatment for prolapse, surgery for prolapse, surgery for fistula, surgery for dehiscence/evisceration. Values of p<0.05 were considered statistically significant.

Results:

In this cohort, 481 women undergoing cystectomy were identified between January 1^st, 2011-December 31^st, 2017. Of these, 179 (37.2%) were less than 70 years old and 424 (88.1%) were white (Table 1). The majority had bladder cancer (N=378, 79%), and were managed with an incontinent conduit or catheterizable channel diversion (N=401, 83.4%). Women undergoing cystectomy for benign conditions (such as interstitial cystitis) were found to be significantly younger as compared to those treated for bladder or other cancer (median: 61 vs. 73 and 70 respectively, p<0.0001). Few patients (N=17, 3.5%) had a history of radiation or underwent robotic cystectomy (N=39, 8.1%).

Table 1.

Baseline characteristics of Radical Cystectomy cohort from 2011-2017, by indication

	Overall	Bladder cancer	Other cancer	Benign conditions	p-value
N	481	378	54	49
Age group, %					0.0001
<70	179 (37.2)	121 (32.0)	26 (48.1)	32 (65.3)
70-74	115 (23.9)	93 (24.6)	—	—
75-79	97 (20.2)	82 (21.7)	—	—
80+	90 (18.7)	82 (21.7)	—	—
Race, %					0.2740
White	424 (88.1)	338 (89.4)	46 (85.2)	40 (81.6)
Black	40 (8.3)	30 (7.9)	—	—
Other/unknown	17 (3.5)	—	—	—
Smoking status, %					0.0213
Never	254 (52.8)	187 (49.5)	39 (72.2)	28 (57.1)
Former	158 (32.8)	131 (34.7)	—	—
Current	69 (14.3)	60 (15.9)	—	—
Obesity, %	74 (15.4)	50 (13.2)	11 (20.4)	13 (26.5)	0.0293
Charlson comorbidity index (median [25th, 75th])	6 (3, 9)	5 (3, 9)	9 (8, 10)	4 (2, 6)	<.0001
US Census Region, %					0.0479
South	170 (35.3)	134 (35.4)	12 (22.2)	24 (49.0)
Midwest	116 (24.1)	86 (22.8)	16 (29.6)	14 (28.6)
West	102 (21.2)	80 (21.2)	—	—
Northeast	93 (19.3)	78 (20.6)	—	—
Diversion type, %					<.0001
Incontinent conduit or catheterizable channel	401 (83.4)	338 (89.4)	24 (44.4)	39 (79.6)
Neobladder	—	—	—	—
Ureterosigmoidostomy or ureterocutaneous diversion	—	—	—	—
Other	52 (10.8)	—	29 (53.7)	—
Radiation history*, %	17 (3.5)	—	—	0 (0.0)	<.0001
Robotic surgery*, %	39 (8.1)	39 (10.3)	0 (0.0)	0 (0.0)	<.0001

Data not reported in cells with long dash (—) to comply with the cell size suppression policy of the Centers for Medicare and Medicaid Services, which requires that no cell can be reported if it allows a value of 1 to 10 to be derived (see https://www.resdac.org/articles/cms-cell-size-suppression-policy)

^*P-values from Fischer’s exact test due to presence of zero cell values

Within two years of cystectomy, 93 patients (19.5% of the cohort) had one or more vaginal complications on record (Figure 1). 78 women had one complication, 15 women had 2 complications, and 388 women had no vaginal complications. Vaginal complications were identified by ICD codes in 117 patients (codes listed in Supplementary Table 1), by CPT codes in 38 patients (codes listed in Supplementary Table 2), and 326 women had a combination of both codes. Vaginal cuff dehiscence had the highest cumulative incidence, occurring in 49 patients (10.2%). This complication also occurred earliest in the postoperative timeline. The median [25^th, 75^th] time to vaginal cuff dehiscence within 2 years from cystectomy was 25 days [11, 62]. Vaginal fistula developed in 33 (6.9%) within the first 2 years after cystectomy. Pelvic organ prolapse in the first 2 years after cystectomy was found in 4.5% (21 women). Relatively few women (< 11) had a diagnosis of dyspareunia over the 2 years following cystectomy.

Figure 1:

Cumulative Incidence Rate of Vaginal Complications.

In both univariate and multivariable cause-specific Cox proportional hazards model (Table 2), age at diagnosis was associated with a 3% decrease in risk (i.e. hazard) of developing vaginal complications relative to a 1-year increase in age (aHR 0.97 95% CI: 0.95-0.99). Similarly, in both unadjusted and adjusted models, surgical indication was associated with vaginal complications. Risk of developing a vaginal complication was 2.33-fold higher in women undergoing cystectomy for non-urologic cancers compared to women undergoing radical cystectomy for bladder cancer, while holding other covariates constant (aHR 2.33 95% CI: 1.17, 4.63). Other covariates were not significantly associated with development of vaginal complications in the adjusted model. Despite relatively high death rates (30.98%; N=149) in this cohort within 2 years of follow-up after index cystectomy, direction and significance of reported associations remained consistent in the multivariable Fine and Gray sub-distribution model that explicitly accounted for death as a competing event (Table 3). 239 women (49.7% of total cohort) survived complication free.

Table 2:

Associations between Covariates and Vaginal Complications during 2-year follow-up: Vaginal Complications Specific Cox Proportional Hazards Regression Model.

Variable	Unadjusted Hazard Ratio (95% CI)	p-value	Adjusted Hazard Ratio (95% CI)	p-value
Age	0.97 (0.95,0.98)	< .001	0.97 (0.95,0.99)	.009
Race
White	1.00 [Reference]		1.00 [Reference]
Black	0.76 (0.33,1.73)	.51	0.64 (0.27,1.49)	.30
Other/unknown	1.05 (0.33,3.34)	.93	0.98 (0.30,3.22)	.98
Charlson comorbidity index	1.04 (0.99,1.10)	.14	1.04 (0.97,1.11)	.24
Current smoking	1.26 (0.79,2.02)	.34	1.17 (0.71,1.95)	.54
Obesity	1.00 (0.57,1.77)	.99	0.85 (0.46,1.55)	.59
Radiation history	1.88 (0.76,4.62)	.17	1.12 (0.42,2.97)	.82
Diversion type
Incontinent conduit or catheterizable channel	1.00 [Reference]		1.00 [Reference]
Neobladder	1.45 (0.59,3.61)	.42	1.36 (0.54,3.45)	.51
Other	2.22 (1.29,3.82)	.004	1.34 (0.67,2.67)	.41
Ureterosigmoidostomy or ureterocutaneous diversion	2.29 (0.72,7.28)	.16	1.31 (0.40,4.36)	.66
Indication
Bladder cancer	1.00 [Reference]		1.00 [Reference]
Benign conditions	2.22 (1.25,3.91)	.006	1.56 (0.77,3.15)	.21
Other cancers	3.13 (1.90,5.16)	< .001	2.33 (1.17,4.63)	.02

Table 3.

Associations between Covariates and Vaginal Complications during 2 year followup: Fine and Gray Sub-Distribution Hazards Model.

Variable	Unadjusted Hazard Ratio (95% CI)	p-value	Adjusted Hazard Ratio (95% CI)	p-value
Age	0.97 (0.95,0.98)	< .001	0.97 (0.95,0.99)	.002
Race
White	1.00 [Reference]		1.00 [Reference]
Black	0.75 (0.33,1.70)	.49	0.62 (0.27,1.39)	.24
Other/unknown	0.86 (0.28,2.64)	.79	0.68 (0.23,1.99)	.49
Charlson comorbidity index	1.02 (0.97,1.08)	.37	1.02 (0.96,1.09)	.46
Current smoking	1.26 (0.79,2.01)	.33	1.16 (0.73,1.85)	.54
Obesity	0.98 (0.56,1.73)	.95	0.81 (0.46,1.44)	.48
Radiation history	1.61 (0.68,3.81)	.28	0.92 (0.35,2.39)	.86
Diversion type
Incontinent conduit or catheterizable channel	1.00 [Reference]		1.00 [Reference]
Neobladder	1.47 (0.61,3.53)	.39	1.31 (0.59,2.95)	.51
Other	1.99 (1.14,3.46)	.01	1.19 (0.57,2.48)	.64
Ureterosigmoidostomy or ureterocutaneous diversion	2.23 (0.79,6.31)	.13	1.29 (0.42,4.01)	.66
Indication
Bladder cancer	1.00 [Reference]		1.00 [Reference]
Benign conditions	2.27 (1.28,4.03)	.005	1.50 (0.74,3.05)	.27
Other cancers	2.98 (1.82,4.90)	< .001	2.55 (1.22,5.31)	.01

Within the first 2 years after cystectomy surgery, 26 (5.4%) women underwent an intervention for management of a vaginal complication following cystectomy. Of these, 12 (2.5%) women had a surgical intervention for management of vaginal prolapse. Relatively few women had interventions for vaginal fistula (< 11) or vaginal cuff dehiscence (<11). Over the entire study period (2011 – 2017), 102 women were diagnosed with a vaginal complication, and 27 received an intervention after the index surgery.

Discussion:

Among women who undergo cystectomy, vaginal complications occur at rates higher than expected and about 5% of women had an intervention for a vaginal complication in the first 2 years after surgery. Radical cystectomy (RC) is the standard treatment of non-metastatic, muscle invasive bladder cancer or high-risk non-muscle invasive bladder cancer. For women, this procedure has traditionally included removal of the anterior vagina, uterus, fallopian tubes, and ovaries, in addition to the bladder and urethra. Improvement in life expectancy after radical cystectomy (RC) for bladder cancer has allowed for increased focus on issues related to cancer survivorship.^2,3 This has resulted in modifications to the standard RC technique with a goal of preserving reproductive and sexual function. However, radical cystectomy with a vaginal sparing or genital organ sparing (vagina/uterus/ovaries/fallopian tubes) approach may be associated with complications specific to women: pelvic organ prolapse, vaginal dehiscence or evisceration, vaginal fistula, and dyspareunia. This paper is the first to comprehensively assess these risks using a large national database.

We found that about 7% of women developed a vaginal fistula within the first 2 years after cystectomy. This could represent a fistula between the vaginal wall and the neobladder, the peritoneal cavity, or the bowel. In a systematic review of vaginal complications after radical cystectomy for bladder cancer, Richter et al. reported that the most common type of vaginal fistula reported is in women undergoing orthotopic diversions occurring between the neobladder and vaginal wall, with reported incidence up to 33% but more consistently ranging from 3-6% at higher volume centers.⁴ The location of neobladder vaginal fistulas most often occurs along the anterior vaginal wall at the location of the neobladder-urethral anastomosis.⁵ The fistula rate reported in our study (7%) likely reflects the broad range of surgical locations within the Medicare LDS, not just high-volume centers. Overall, relatively few patients in our cohort had any type of continent diversion (over 80% of our cohort had an incontinent conduit or catheterizable channel diversion). Given that our Medicare cohort was an older group, it makes sense that they were less likely to have a continent diversion, as this is often reserved for younger, healthy patients. Understanding that most fistula occur in the setting of neobladder reconstruction, it is plausible that the incidence of vaginal fistula in the general population is even higher than what we found in an older cohort with very few neobladder diversions.

We found the most common vaginal complication after cystectomy to be vaginal cuff dehiscence, occurring in about 10% of the cohort. This complication is of particular concern as it can involve bowel evisceration and the need for emergent reoperation. In a multicenter study, Kanno et al. in 2019 reported on 100 women undergoing laparoscopic radical cystectomy, seven of whom underwent emergent reoperation for vaginal dehiscence with bowel evisceration.⁶ The authors hypothesized that surgical factors (length of anterior vaginal wall resection, surgical approach, excessive use of electrocautery) and patient factors (age and vaginal atrophy) may be contributors.⁶ Although the data are limited, minimally invasive approaches to hysterectomy (such as laparoscopic or robotic hysterectomy) seem to be associated with a higher risk of vaginal cuff dehiscence.⁷ Unfortunately, because of the low numbers of robotic cystectomy in our cohort, the impact of minimally invasive cystectomy on rates of vaginal dehiscence could not be determined. In the gynecologic literature, the incidence of vaginal dehiscence after pelvic surgery (specifically hysterectomy) is described from <1 to 4%, with the mean time to cuff dehiscence varying from 6.1 weeks to 1.6 years (range, 2 weeks to 5.4 years).⁷ As depicted in Figure 1, vaginal evisceration was the earliest occurring vaginal complication in our cohort, with the majority of cases presenting within the early postoperative months, and almost all cases presenting before 1 year postoperatively.

The overall rates of dyspareunia were low, occurring in <11 women. This is likely due to a variety of factors, including reporting bias, age of cohort, and method of data collection. The Medicare dataset does not indicate whether women are sexually active after cystectomy, which is an important survivorship issue, particularly in women who undergo vaginal sparing procedures for the purpose of maintaining ability for intercourse.

Other research has described individual complications after radical cystectomy but this study provides a comprehensive assessment of vaginal complications after cystectomy using a national database. The study is strengthened by a large sample size and the assessment of risk factors influencing complication rates. The study is limited by biases inherent to billing data. Use of diagnosis and procedure codes may have underestimated the true prevalence of vaginal complications in the cystectomy population. Coding data may also be at risk of mischaracterizing diagnoses that are difficult to differentiate clinically, such as “enterocele” (where vaginal epithelium is overlying protruding small bowel) as compared to true vaginal cuff dehiscence with bowel evisceration. In addition, we were unable to assess how specific aspects of surgical technique (sparing of uterus, sparing of vaginal wall) influenced outcomes. Nor were we able to assess whether surgical modifications, such as interposition flaps or minimally invasive surgical techniques, influenced the development of complications. Furthermore, clinical details such as symptom duration and level of bother are not included in the analysis, thereby limiting any comment on quality of life impact.

Overall, we found that approximately 20% of women who had a cystectomy between 2011 and 2017 in the Medicare 5% sample had a vaginal complication. The complication rate was higher than expected, especially since billing/coding databases typically underestimate true prevalence. Moreover, we found that 31% of women died within 2 years of surgery, leaving approximately 50% alive and free of vaginal complications within 2 years of the surgery. Urologic oncologists should discuss these potential complications with patients preoperatively, and should provide them with information about possible postoperative symptoms. These findings also highlight a need for prospective studies, utilizing standardized instruments and subjective outcome measures.

Conclusions:

Approximately 20% of women in our Medicare 5% sample had a vaginal complication following their cystectomy. Further studies, utilizing standardized instruments and subjective outcome measures are needed to determine the impact of vaginal complications on quality of life.

Supplementary Material

Supplementary Table 2

Supplementary Table 2: CPT codes for surgical and non-surgical management of pelvic organ prolapse, vaginal fistula, and vaginal dehiscence or evisceration.

Supplementary Table 1

Supplementary Table 1: ICD-9 and ICD-10 codes used to identify patients with pelvic organ prolapse, vaginal fistula, dyspareunia, vaginal dehiscence or evisceration.