Vesicoureteral Injury during Benign Hysterectomy: Minimally Invasive Laparoscopic Surgery versus Laparotomy

There were 509 395 women who had undergone hysterectomies for benign gynecologic disease. Of those, 8285 women with an unknown hysterectomy type or who had undergone radical hysterectomy were excluded, and the remaining 501 110 women represented the study population. The most common hysterectomy type was TAH (n = 284 365 [56.7%]), followed by TLH (n = 60 410 [12.1%]), Abd-SCH (n = 55 655 [11.1%]), LAVH (n = 45 620 [9.1%]), TVH (n = 34 865 [7.0%]), and LSC-SCH (n = 20 195 [4.0%]). Patient demographics are shown in Table 1 . All of the examined covariates differed significantly across the hysterectomy modes (all, p <.001). Specifically, women who had undergone TVH were older than those who had undergone other hysterectomy types, and those who had Abd-SCH were more likely to be black, have class III obesity, and have Medicaid. Women who had undergone TLH were more likely to have higher comorbidity indices and to have undergone surgery at a large or urban teaching hospital. Those who had undergone LSC-SCH were more likely to have a higher median household income. Women in the TAH group were more likely to have undergone surgery in the Southern region of the United States and to have undergone concurrent oophorectomy. The characteristics of gynecologic disease are shown in Table 1 . Women in the LAVH group were more likely to have endometriosis or abnormal uterine bleeding, whereas those in the Abd-SCH group were more likely to have uterine leiomyomas or pelvic infection. Women in the TAH group were more likely to have adnexal pathology, and those in the TVH group were more likely to have a uterine polyp. Vesicoureteral injury rates overall (either ureteral or bladder injury) were assessed by hysterectomy type ( Fig. 1 ). In the whole cohort, vesicoureteral injury was reported in 1045 (0.21%) women. Vesicoureteral injury (either ureteral or bladder injury) rates differed statistically significantly across the hysterectomy types (p<.001); LAVH (0.28%; 130/45 620) had the highest bladder injury rate, followed by TVH (0.24%; 85/34 865) and TLH (0.24%; 145/60 410), TAH (0.20%; 580/284 365), Abd-SCH (0.15%; 85/55 655), and, finally, LSC-SCH (0.10%; 20/20 195). When vesicoureteral injury was further stratified by location of injury, LAVH (0.23%) had the highest rate of bladder injury, followed by TVH (0.20%), TAH (0.16%), Abd-SCH (0.12%), TLH (0.12%), and LSC-SCH (0.05%) (p <.001; Fig. 1 ). For ureteral injury rate, TLH (0.13%) had the highest rate, followed by TVH (0.06%), LAVH (0.06%), LSC-SCH (0.05%), Abd-SCH (0.04%), and TAH (0.04%) (p <.001; Fig. 1 ). Patient demographics and gynecologic disease characteristics between the TLH and TAH groups were balanced in the PS-IPTW model ( Supplemental Table S2 ). All 17 covariates were well balanced between the 2 groups (all, SD ≤.10), and 14 724 TLH cases and 11 410 TAH cases were assessed for vesicoureteral injury ( Fig. 2 ). TLH was associated with an increased risk of vesicoureteral injury overall compared with TAH (0.31% vs 0.20%, OR 1.57, 95% CI 1.01–2.42, p = .045). When injury sites were further stratified, compared with TAH, TLH was significantly associated with an increased risk of ureteral injury (0.16% vs 0.04%, OR 3.95, 95% CI 2.03–7.67, p <.001) but not bladder injury (0.17% vs 0.17%, OR 1.04, 95% CI 0.57–1.90, p = .897). When the risk of ureteral injury was further stratified on the basis of the gynecologic disease type ( Table 2 ), the presence of endometriosis had the highest impact on ureteral injury for TLH compared with TAH (OR 6.15, 95% CI 1.18–31.9, p = .031), followed by the presence of uterine myoma (OR 4.15, 95% CI 2.13–8.11, p .05). Among TLH cases, robotic-assisted and conventional approaches were compared ( Supplemental Table S3 ), and there were no differences in vesicoureteral injury (OR 0.59, 95% CI 0.28–1.24. p = .162), including ureteral injury (OR 0.72, 95% CI 0.26–1.94, p = .510), between the 2 groups. For the second comparison, vesicoureteral injury was compared between the Abd-SCH and LSC-SCH groups. Differences in patient demographics and gynecologic disease types between the 2 groups were balanced in the PS-IPTW model ( Supplemental Table S4 ). All of the tested covariates were well balanced between the 2 groups (all, SD ≤.10), and 5211 LSC-SCH cases and 11 593 Abd-SCH cases were assessed for vesicoureteral injury ( Fig. 3 ). LSC-SCH was not associated with an increased risk of vesicoureteral injury when compared with Abd-SCH (0.08% vs 0.14%, OR 0.62, 95% CI 0.19–1.98, p = .419). When injury types were specified, LSC-SCH was not associated with an increased risk of bladder injury (0.06% versus 0.10%, OR 0.50, 95% CI 0.10–2.36, p = .377) and ureteral injury (0.04% vs 0.04%, OR 0.82, 95% CI 0.15–4.48, p = .822). Similarly, LAVH was not associated with an increased risk of vesicoureteral injury compared to TVH ( Supplemental Table S5 and Fig. 4 ) (0.29% vs 0.24%, OR 1.21, 95% CI 0.63–2.33, p = .564). LAVH was not associated with an increased risk of bladder injury (0.24% vs 0.19%, OR 1.27, 95% CI 0.62–2.59, p = .510) or ureteral injury (0.05% vs 0.06%, OR 0.80, 95% CI 0.19–3.40, p = .761) compared with that associated with TVH. Finally, albeit not statistically significant, the LAVH group had a slightly higher rate of vesicoureteral injury than did the TAH group (0.31% vs 0.20%, OR 1.55, 95% CI 0.97–2.45, p = .065; Supplemental Table S6 ).

The National Inpatient Sample (NIS) is a publicly available and de-identified population-based database that is distributed as part of the Healthcare Cost and Utilization Project by the Agency for Healthcare Research and Quality [ 7 ]. The NIS database includes hospital discharge data for more than 36 million hospitalizations per year when weighted, and it provides information on patient characteristics and resource use, such as diagnosis and intervention types, length of stay, and hospital charges, in addition to hospital-specific data, including location, bed size, and teaching status. More than 90% of the US population is represented in this database during the study period when weighted. The University of Southern California Institutional Review Board deemed the study exempt owing to the use of publicly available de-identified data. This is a retrospective observational study examining the NIS database between January 2012 and September 2015. Women who had undergone inpatient hysterectomy for benign gynecologic disease were eligible for the study. Women who had undergone a hysterectomy for an oncological indication or an unknown indication were excluded. Women for whom there was no information on hysterectomy mode or who had undergone a radical hysterectomy were also excluded. Among cases eligible for analysis, the following information was abstracted from the NIS database: patient demographics, disease factors, hospital information, operative types for hysterectomy, and outcome of the index admission. Patient demographics included age, race/ethnicity, medical comorbidities, obesity, primary expected payer, and median household income. The Charlson Comorbidity Index was determined for each patient on the basis of the codes for the specified medical conditions in each category and weighted appropriately to calculate a final score [ 8 ]. Disease factors included benign uterine and adnexal pathology. Gynecologic disease information included the presence of uterine myomata, endometriosis, abnormal uterine bleeding, pelvic infection, uterine polyp, and/or adnexal pathology. Surgical approaches were divided into the following groups: total abdominal hysterectomy (TAH), total laparoscopic hysterectomy (TLH), abdominal supracervical hysterectomy (Abd-SCH), laparoscopic supracervical hysterectomy (LSC-SCH), total vaginal hysterectomy (TVH), and laparoscopy-assisted vaginal hysterectomy (LAVH). Performance of oophorectomy was also recorded. Hospital data included the calendar year of hospitalization, hospital bed size, teaching status, and hospital region. Hospital bed size is defined by hospital geographic region, urban–rural designation, and teaching status [ 9 ]. Operative details included mode of hysterectomy (laparotomy, minimally invasive laparoscopic, and vaginal) and type of hysterectomy (total simple or supracervical). The outcome for analysis was perioperative vesicoureteral injury diagnosed during the index admission. During the study period from January 2012 to September 2015, the International Classification of Diseases, Ninth Revision codes remained the same ( Supplemental Table S1 ). Obesity was grouped as class I-II obesity (body mass index, 30–39.9 kg/m 2 ) and class III obesity (body mass index, ≥40 kg/m 2 ). The International Classification of Diseases, Ninth Revision codes for bladder injury (8670 and 8671) and ureteral injury (8672 and 8673) were used to designate the presence of a vesicoureteral injury [ 10 ]. In the NIS database, perioperative complications during the index admission for hysterectomy were recorded; there is no differentiation between intraoperative and postoperative events, and there is no information on complications that may have occurred after discharge. The primary objective of the analysis was to examine the rate of vesicoureteral injury by hysterectomy approach among all hysterectomies performed for benign gynecologic disease. The secondary objective was to compare the risk of vesicoureteral injury on the basis of use of a minimally invasive laparoscopic approach compared with laparotomy. Specifically, risk of vesicoureteral injury was assessed in the following 2 cohorts: (1) TLH versus TAH, and (2) LSC-SCH versus Abd-SCH. The fundamental rationale behind this comparison was to assess the risk of vesicoureteral injury on hysterectomy performed through a minimally invasive laparoscopic approach relative to that of laparotomy, which is historically the gold standard. All analyses for the primary objective was based on the weighted values provided by the NIS program. Differences in continuous variables were assessed using the 1-way analysis of variance test. Differences in ordinal and categorical variables were assessed using the χ 2 test. Propensity score (PS)–based inverse probability of treatment weighting (IPTW) was fitted to corroborate the background differences in the 2 groups on comparing TLH versus TAH or Abd-SCH versus LSC-SCH [ 11 ]. First, PS for laparoscopy use was determined by fitting a binary logistic regression model [ 12 ]. All data on patient demographics, gynecologic pathology, and oophorectomy use were entered in the final model. Then, the IPTW approach was used to assign patients who had undergone laparoscopy a weight of 1/PS and those who had undergone laparotomy a weight of 1/(1-PS) [ 11 ]. Stabilized weights were used in the analysis, and the threshold technique was used at the 1st and 99th percentiles of the weight distribution [ 11 ]. In the PS-IPTW model, a proportional distribution of baseline covariates was assessed for effect size, and a standardized difference (SD) of 0.10 or less was considered a good balance between the 2 groups. After the PS-IPTW modeling, a generalized estimating equation model was fitted to estimate the magnitude of statistical significance for vesicoureteral injury, expressed as the odds ratio (OR) and 95% confidence interval (CI). All statistical analyses were based on 2-tailed hypotheses, and a p-value of less than 0.05 was considered statistically significant. SPSS software version 24.0 (IBM Corp., Armonk, NY) was used for all analyses. The Strengthening the Reporting of Observational Studies in Epidemiology guidelines were used to present the performance of the observational study [ 13 ]. Various sensitivity analyses were undertaken to assess the robustness of the study findings. First, the risk of vesicoureteral injury was compared between the specific modes of minimally invasive laparoscopy: conventional laparoscopy versus robotic-assisted laparoscopy. This comparison was performed in the total hysterectomy cohort. Second, the risk of vesicoureteral injury was assessed on the basis of the type of benign gynecologic disease. The rationale behind this analysis was that the surgical complexity would be different across the gynecologic disease types. Third, the vesicoureteral injury risk was assessed between the LAVH and TVH groups. Finally, the vesicoureteral injury risk was compared between the LAVH and TAH groups. This comparison was based on post hoc observation of the highest vesicoureteral injury rate in the LAVH group.

Although TLH is associated with a nearly 4-fold increased risk of ureteral injury but no difference in bladder injury when compared with TAH, the incidence of vesicoureteral injury overall is low for all routes of hysterectomy. Preoperatively, individualized counseling of all patients considering TLH should be performed, balancing the benefits of laparoscopy, such as shorter recovery time and hospital stay, with the heightened risk of ureteral injury as well as taking into consideration any other patient-specific risk factors for ureteral injury. This can be particularly applicable for certain gynecologic diseases such as endometriosis, as this condition exhibited the highest impact on ureteral injury with TLH.

Our findings are consistent with the reported rates of urinary tract injury in previous publications [ 14 – 17 ]. The aforementioned meta-analysis reported a urinary tract injury rate of 0.24% and 0.1% for laparoscopic and abdominal hysterectomies, respectively, among 2140 women (OR 2.44) [ 5 ]. Another study of 876 women, encompassing 584 laparoscopic and 292 abdominal hysterectomies, reported an increase in ureteral injury rates in laparoscopic compared with abdominal hysterectomy cases (0.9% vs 0%) [ 18 ]. In an English study in 2015 that examined 310 105 hysterectomies for benign conditions, the rate of ureteric injury has been reported as 0.6% for laparoscopic hysterectomies compared with 0.3% for the open approach [ 14 ]. A systematic review from 2014 reported an overall urinary tract injury rate of 0.73%, with a specific ureteral injury rate of 0.02% to 0.4%, and concluded that laparoscopic hysterectomy did not pose a higher risk than that of the abdominal approach; however, the studies included had smaller sample sizes and were likely underpowered to detect a statistically significant difference [ 19 ]. Finally, a Finnish study that examined 5279 hysterectomies reported a 0.3% risk of ureteral injury with laparoscopic hysterectomy; however, the rates of vesicoureteral injury were similar to those of abdominal hysterectomy [ 20 ]. It is important to note that their study did not further stratify cases by route of hysterectomy and likely included supracervical approaches. In our study, risk of ureteral injury was higher for TLH than for TAH; however, this result was not observed when examining LSC-SCH versus Abd-SCH in addition to LAVH versus TVH. Ureteral injury commonly occurs at the level of the pelvic brim and cardinal ligaments and before insertion of the cardinal ligaments into the bladder; thus, we postulate that ureteral injury may occur during dissection of the cardinal ligaments or while closing the vaginal cuff during TLH [ 16 , 21 ]. Although supracervical hysterectomy also involves the aforementioned steps with the exception of cuff closure, dissection is required only to the level of the mid cervix and may minimize iatrogenic injury to the bladder and/or ureter. There are several strengths to this study. First, given that vesicoureteral injuries are uncommon among all types of hysterectomy, use of a population-based national database allowed for a large enough sample size to detect significant differences in ureteral injury rates. In addition, background corroboration with PS-IPTW enhanced our statistical rigor given that patients may be at different risks of ureteral injury on the basis of their demographics and clinical variables. This method was also particularly useful as the rate of vesicoureteral injury regardless of hysterectomy type overall was low and other background adjustments such as PS matching would have limited our study size and lowered statistical power. Moreover, hysterectomy mode–specific analysis (total hysterectomy, supracervical hysterectomy, and vaginal) clearly highlighted the difference in vesicoureteral injury risk attributed to minimally invasive laparoscopic approaches. Our study has several limitations. First, although the association between ureteral injury and TLH was robust when compared with TAH, there may be additional variables that contribute to risk of ureteral injury, such as parity, uterine size, surgical history (specifically of Cesarean delivery), surgical site adhesions, extent of endometriosis, surgeon’s experience and skills, and surgical difficulty, that were not captured in this study because of the limitations associated with a populational database. Second, owing to the nature of the NIS, which was designed for administrative purposes rather than solely specific research purposes, coding misclassifications or omissions may have led to a bias in the results. Unless the archived medical records are retrieved for review, the accuracy of data entry and the specific preoperative indication for hysterectomy remain unknown. The risk of vesicoureteral injury likely differs depending on the hysterectomy indication. The study team acknowledged this as a major drawback of this study. In addition, this database only includes discharge data from a single admission, thus it is unable to account for occult injuries that were not identified immediately intraoperatively or during the admission. Given that previous studies have cited that up to 87% of ureteral injuries may be delayed, our study may underestimate the true number of urinary tract injuries [ 22 ]. However, this underestimation likely applies to all hysterectomy groups (nondifferential outcome misclassification). Finally, the NIS database captures inpatient admissions only, and lack of information on same-day hysterectomies results in selection bias, given the increasing number of same-day minimally invasive hysterectomies in the United States. In our dataset, only one-third of the patients had undergone minimally invasive hysterectomy between 2012 and 2015; this clearly reflects the omission of a substantial number of same-day minimally invasive hysterectomies, as nearly half of the hysterectomies were performed through a minimally invasive approach in 2010 [ 3 ]. If same-day hysterectomy is presumed to (1) be more commonly performed through a laparoscopic approach and (2) have lower complication rates than those of inpatient hysterectomy, it is speculated that the true incidence of vesicoureteral injury at the time of minimally invasive laparoscopic hysterectomy would likely be lower than that reported in this study. Given the findings of our study, consideration should be given to specifically discussing the increased risk of ureteral injury during the process of obtaining informed consent for minimally invasive hysterectomy. In addition, consideration of utilization of supracervical hysterectomy in appropriate candidates with normal cervical cancer screening results is reasonable given the elevated risk of ureteral injury with total compared with supracervical hysterectomy. In the case of endometriosis, particularly a severe type as identified during laparoscopy, early decision for laparotomy conversion would be key to avoiding possible risks of ureteral injury.