Risk factors for vaginal cuff dehiscence after robot-assisted total laparoscopic hysterectomy: A retrospective cohort study.

Silje Eilertsen Denstad: Conceptualization, project administration, data curation, investigation, formal analysis, writing – original draft. Ingrid Volløyhaug: Conceptualization, supervision, writing – review and editing. Marit Lieng: Conceptualization, supervision, writing – review and editing. Gaby Moawad: Conceptualization (suggested risk factors), writing – review and editing. Risa Anna Margaretha Lonnee‐Hoffmann: Conceptualization, supervision, project administration, writing – review and editing.

The study has been approved by the Regional Committees for Medical and Health Research Ethics (REK‐Midt, reference number: 2020/155169), on September 8, 2020 with amendment approved March 30, 2023. The study has been reviewed and approved by the Data Protection Officer at St. Olav's Hospital and a Data Protection Impact Assessment and a risk and vulnerability analysis have been conducted. The study was registered at the Open Science Framework (OSF) with the following link: https://osf.io/hs7gd (Date registered: June 6, 2022).

The Foundation Dam through the Norwegian Women's Public Health Association (Norske Kvinners Sanitetsforening) funded this study.

The final study population comprised 684 women who underwent RA‐TLH at Trondheim University Hospital during the study period, representing 44.2% of 1549 eligible women (Figure  1 ). Flowchart for inclusion of participants in the study. Their median age was 47 years (range 25–83) at the time of hysterectomy (mean: 48.9, SD: 10.9). Their median BMI was 26.5 kg/m 2 (range 17.0–51.2) (mean: 27.4, SD: 5.3). Among the 552 women who reported their relationship status, 465 (84%) had a partner prior to surgery. Of the 539 women who reported their smoking status, 110 (20%) were smokers. Indications for surgery included bleeding disorders ( n  = 207, 30%), cancer ( n  = 175, 26%), premalignant conditions ( n  = 129, 19%), myoma ( n  = 118, 17%), endometriosis or pain ( n  = 76, 11%), and other indications ( n  = 59, 9%). The median weight of the uterine specimen was 150 grams (range 35–950) (mean: 195, SD: 141), and the median duration of surgery was 99 minutes (range 43–318) (mean: 107, SD: 37). During the review of medical records, we identified 28 (4.2%) women with VCD. Additionally, 12 women were identified with typical symptoms consistent with VCD, but they were examined with several days to months delay, and no definitive diagnosis of VCD was made. This occurred more commonly in the early years following the introduction of RA‐TLH (Table  1 ). According to the definition of VCD in this study, these women were not included as cases. Excluding these not confirmed VCD cases from the non‐VCD cases did not change the results of the analysis. Figure  2 shows the trends in average uterine weight, duration of surgery, and the proportion of confirmed VCD during the study period. Cases of confirmed and suspected VCD after RA‐TLH by time period. Trends in uterus weight (g), length of surgery (min), and confirmed vaginal cuff dehiscence (VCD, %) during the study period. For women with VCD, vaginal intercourse was the triggering event in 23 women (82.1%). The median time interval between surgery and VCD was 73 days (range 28–278) and mean 98 days (SD 59), with 25 (89.3%) occurring within the first 6 months post‐hysterectomy. The most commonly reported symptoms were acute abdominal pain (83.9%, n  = 25) and vaginal bleeding (46.4%, n  = 13). Additional symptoms included vaginal discharge (21.4%, n  = 6), sensation of a lump in the vagina (17.9%, n  = 5), and fever (14.3%, n  = 4). In 16 women (57.1%), more than 50% of the vaginal cuff was affected by wound dehiscence, while in seven women (25.0%), less than 50% was affected. The extent of the rupture was not documented in five women (17.9%). Seven women (25.0%) experienced bowel evisceration into the vagina, and three women (10.7%) had recurrent ruptures. The results of the univariable logistic regression analysis are shown in Table  2 . Women who experienced VCD were significantly younger and had lower BMI compared to women without VCD. They had significantly longer duration of surgery and higher uterine weight compared to women without VCD, but no significant differences for other demographic variables or factors related to surgery were found; see Table  2 . No consistent trend in VCD occurrence was observed across individual surgeons based on procedure volume or duration of experience with RA‐TLH. Demographic, surgical, and postoperative factors of women with and without vaginal cuff dehiscence (VCD) after robot‐assisted total laparoscopic hysterectomy. Surgery score: 0 = no prior abdominal surgery; 1 = laparoscopy; 2 = open. In the multivariable logistic regression analysis (Table  3 ), we found that each unit increase in BMI was associated with decreased odds with aOR 0.87 (95% CI 0.78–0.97) and each 100 grams increase in uterine weight increased the risk for VCD with aOR 1.3 (95% CI 1.1–1.6). Logistic regression analysis of risk factors for vaginal cuff dehiscence.

In this study, we aimed to investigate patient‐ and surgery‐related risk factors for VCD following RA‐TLH. We found that vaginal intercourse was the most common trigger for VCD. Increasing BMI was a protective factor, whereas higher uterine weight was associated with increased risk of VCD in multivariate analysis. In our cohort, the VCD rate following RA‐TLH was 4.2%. This is high compared to the pooled incidence rate of 1.7% after RA‐TLH reported in a recent systematic review and meta‐analysis. 13 The authors emphasized that methodological heterogeneity, such as differences in surgical technique, study population, and follow‐up protocols, makes it difficult to establish a reliable incidence rate across studies ranging from 0.4% to 4.1%. 6 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 Several factors may explain the discrepancy between our findings and those of studies reporting lower incidence rates. Two studies with rates below 1% included older women with higher BMI, both of which were identified as protective factors for VCD in our analysis. 16 , 22 Another study with an incidence below 1% relied on ICD coding for diagnosis, which may underestimate true incidence due to miscoding and missed cases. 20 By contrast, studies with study populations more comparable to ours reported rates of VCD of 3.2% and 2.6%. 5 , 15 The duration of postoperative follow‐up may also contribute to the observed differences. Our 12‐month follow‐up period exceeds the typical 6 months in most studies. 16 , 18 , 22 When limiting our analysis to VCD cases diagnosed within the first 6 months, our incidence was 3.7%. We acknowledge that unmeasured factors, such as surgical technique and institutional diagnostic routines, may have influenced the observed incidence. We used a single horizontal continuous layer of V‐Loc barbed sutures for vaginal cuff closure. Although some studies suggest that barbed sutures may reduce the VCD risk, a recent meta‐analysis found no significant difference compared with conventional sutures. 23 Few studies focus specifically on barbed suture closure in robotic procedures, which differ from conventional laparoscopy by lack of tactile feedback. This may potentially affect suture tension and VCD risk. 13 , 23 Lastly, some degree of selection bias is accepted, as participants were informed that the study focused on postoperative complications. Prior research indicates that individuals are more likely to engage in studies they perceive as personally relevant, potentially leading to overrepresentation of women with VCD among the responders. 24 In the early years following the introduction of robotic hysterectomy, several women presented with symptoms indicative of VCD. However, due to delayed examination, definitive diagnosis of VCD was often not established, introducing the possibility of spontaneous healing. Increased awareness in recent years may have led to earlier and more frequent diagnoses of this complication. A systematic review and meta‐analysis from 2021 reported that most of the studies on the etiology of VCD identified intercourse as the precipitating factor. The mean time interval was 68.5 days after hysterectomy. 21 These findings are consistent with our study, with a median interval of 73 days between surgery and VCD, aligning with our hospital's recommendation of 8 weeks of postoperative abstinence. A review postulates that wound strength reaches 30% of its final strength by 3 weeks and approximately 80% after 3 months, but never fully returns to pre‐surgery level. This may indicate that a longer abstinence could be beneficial. 25 The finding that higher BMI correlated with reduced risk of VCD is corroborated by the majority of studies. 19 , 26 , 27 , 28 Donellan et al. hypothesized that obesity might protect against dehiscence due to both structural and physiological properties. 26 They suggested that obese women might be more protected against VCD because of decreased physical force at the vaginal apex during intercourse due to different positions and angles. Additionally, they proposed that a high BMI could be protective because adipose tissue increases impedance in the monopolar circuit, potentially leading to less tissue damage during colpotomy and thereby reducing VCD risk. 26 Uccacella et al. found no protective effect of obesity, most likely because of low prevalence of obese women in their study. 29 In contrast to our results, a South Korean study found that obese women had a 2.2‐fold higher risk of VCD compared with normal weight women, conflicting with most other studies. 30 Differences may be due to study design. In our study, younger age increased risk for VCD in univariate analyses, but the effect of age disappeared in the multivariate analysis, which is in line with Kim et al.'s study. 31 Other studies have identified younger age as a risk factor for VCD. 5 , 7 , 26 , 27 Given that coitus is a prevalent trigger for vaginal cuff rupture, it is plausible that older women exhibit a reduced risk due to less sexual activity. Several authors have argued that the risk factors for VCD differ between premenopausal and postmenopausal women. Vaginal coitus before complete healing of the vaginal cuff is the main trigger of VCD in younger individuals, while vaginal atrophy and pelvic floor prolapse may be more relevant in older women. 32 , 33 , 34 We found that prolonged surgery was a risk factor for VCD in our univariable analysis. This is in line with two other studies. 5 , 35 We suggest that longer duration of surgery may increase the risk of contamination, infection, and tissue trauma, impairing wound healing and elevating the risk of dehiscence. 8 , 10 , 21 , 36 However, several studies did not find this association. 26 , 29 , 31 , 37 Duration of surgery is correlated with uterine weight. We therefore assume that the significant effect of more prolonged surgery in the univariate analysis was mainly caused by larger uteri, implicating more extensive, challenging, and time‐consuming surgery. In contrast to Radosa et al., we found increased risk of VCD with increasing uterine weight in our multiple logistic regression analysis. 37 They attributed their inverse risk to the use of a small‐cup uterine manipulator for smaller uteri, which may leave cervical tissue in the vaginal apex that does not heal well. This issue could also arise with larger uteri, where achieving adequate contact between the cup and the uterus is challenging. Additionally, removing a large uterine specimen via the vagina is often difficult, frequently resulting in traction on the vaginal apex. We suggest that this also increases the risk of VCD. Three other studies found no significant effect of uterine weight on the risk of VCD. 26 , 29 , 38 Cigarette smoking is a well‐known risk factor for wound healing complications. 39 , 40 Constituents of cigarettes cause vasoconstriction and impair the immune response, hindering wound healing. Tobacco also reduces the delivery of essential nutrients for postoperative healing. 39 , 40 Several studies have found an increased risk of vaginal cuff dehiscence in smokers. 18 , 27 , 29 However, we did not find this association, possibly due to some missing data on smoking status and a low proportion of smokers (21.2%). The main strength of this study is the large study population. In addition, we had access to all medical records from public hospitals in our region. The detailed data collection, including a wide range of patient characteristics, preoperative, intraoperative, and postoperative variables, enables a thorough analysis of risk factors. The surgical treatment at the department was relatively uniform. Additionally, the study employs both univariate and multivariate analyses, including DAG analysis, to identify and control for potential confounding factors. By using a DAG, we reduced the risk of bias by controlling for relevant variables. In terms of weaknesses, as a retrospective observational study, there is an inherent risk of selection and information bias. Notably, only 44.2% of the eligible women consented to participate in the study. It is likely that more women who experienced VCD and other complications agreed to participate, as the information sheet indicated that the study focused on risk factors for complications. The missing data for some variables, such as smoking and partner status, may have influenced the results and reduced the statistical power of the study. Furthermore, the absence of data on post‐surgery sexual activity, despite the inclusion of partner status, limited the understanding of this potential risk factor. No routine follow‐up visit was scheduled, and postoperative complications, such as spontaneously resolving VCD, could therefore be underreported, although the clinical relevance of this is minor. Further, the study is limited to a single institution, which may theoretically affect the generalizability of the findings to other settings. However, the surgical approach was following international standards. The clinical implications of our results are that doctors should be aware that women with low BMI and large uteri are at increased risk of VCD after RA‐TLH and inform women at risk. Further studies are needed to establish if, for example, a 3‐month sexual abstinence may reduce the incidence of VCD compared with 8 weeks. Further, removal of a morcellated large uterus through the umbilical port (via an endobag) should be compared with vaginal removal in terms of VCD.

This large study including 684 women identified vaginal intercourse as the predominant trigger for VCD after RA‐TLH. High BMI was a protective factor, whereas large uterine size increased the risk. It is important to inform women with low BMI and a large uterus that they have a potential increased risk of VCD. Future research should focus on determining the optimal duration of sexual abstinence as well as the optimal removal method for a large uterus, following total hysterectomy.

Hysterectomy is one of the most commonly performed gynecological surgical procedures in the Western world. Minimally invasive hysterectomy is preferred over an abdominal approach (laparotomy) when possible because of its advantages, including reduced blood loss, shorter hospital stay, faster recovery, and decreased risk of wound and abdominal wall infections. 1 Over the past two decades, the adoption of minimally invasive hysterectomy techniques, particularly laparoscopic and robot‐assisted methods, has increased significantly in the Western world. 2 , 3 , 4 In 2018, approximately 73% of all hysterectomies in Norway were performed using minimally invasive techniques and approximately 15% of all hysterectomies with robot‐assisted laparoscopic methods. 2 Despite the advantages of minimally invasive hysterectomy, previous studies have indicated an increased risk of vaginal cuff dehiscence (VCD) with laparoscopic and particularly robot‐assisted surgery. 5 , 6 , 7 VCD is a serious complication following total hysterectomy, where the sutured vaginal edges separate postoperatively. 6 , 7 , 8 , 9 , 10 This can lead to peritonitis, sepsis, and bowel injury. Women with VCD often require emergency hospitalization and surgical treatment. 8 The weighted median incidence rates of VCD after different modes of hysterectomy are reported as follows in a recent review: abdominal hysterectomy 0.28%, vaginal hysterectomy 0.5%, laparoscopic hysterectomy 0.87%, and robot‐assisted total laparoscopic hysterectomy (RA‐TLH) 2.33%. 8 Identifying risk factors for VCD is important for enhancing patient safety and optimizing surgical outcomes. Any patient‐related or surgical factors that affect wound healing and tissue quality or increased tension on the vaginal cuff can influence the risk of VCD. 10 Previous studies investigating the impact of patient characteristics on the risk of VCD are limited and have produced inconsistent results. 8 Consequently, it remains unclear which women are at an increased risk for this complication. The aim of this study was therefore to evaluate possible patient and procedure‐related risk factors for VCD in order to provide more specific and knowledge‐based patient counseling.

Silje Eilertsen Denstad is one of the surgeons who performed the procedures on the women discussed in this article. Gaby Moawad is a speaker for Intuitive Surgical. Risa Lonnée‐Hoffmann, Ingrid Volløyhaug, and Marit Lieng declare no conflict of interest.

This is a retrospective observational study of women who underwent RA‐TLH at Trondheim University Hospital, Norway, from November 2010, when RA‐TLH was introduced at the study site, through December 2021. All women undergoing RA‐TLH, both for benign and malignant indications, were invited to participate. Women who were deceased or could not be contacted for consent were excluded. Only hysterectomies where the entire procedure, including colpotomy and cuff closure, was performed robotically were included. All women who had a RA‐TLH during the study period were identified via the hospital's local patient registry. Eligible women received electronic information and consent form via an official website for health services for Norwegian residents, with up to three reminders. Women without registered access to this website received the documents by mail. For consenting participants, we conducted a comprehensive review and systematically extracted data from their electronic medical records. The electronic medical records at Trondheim University Hospital include information from all public hospitals in the region, and no other healthcare institutions in the region handle VCD following RA‐TLH. The variables registered included patient characteristics such as age, body mass index (BMI), parity, having a partner, smoking status, and menopausal status. Previous surgeries were categorized using a surgery score described by Boosz et al.: zero points for no previous surgery, one point for previous laparoscopic surgery, and two points for previous open surgery. 11 Surgical specifications for the RA‐TLH were also recorded, including the name of the surgeon (used to determine months of experience), the indication for surgery, duration of surgery, uterine specimen weight, intraoperative blood loss, and intraoperative complications. Additionally, any postoperative complications within the first year after surgery were registered based on the women's medical records. VCD was defined as a full‐thickness separation of the vaginal incision, with or without the protrusion of abdominal or pelvic organs through the opening. For women with VCD, the timing of diagnosis, any triggering mechanisms, symptoms, and treatments were documented. Although some women may have been treated for postoperative complications elsewhere, only a few belong to other health regions where we do not have direct access to patient records. We included two women who were reoperated for VCD at other facilities, with discharge summaries confirming the reoperations. We reviewed the electronic surgical scheduling system to determine the number of RA‐TLH procedures each surgeon had performed. For each VCD case, we recorded the surgeon's cumulative procedure count at the time of surgery and categorized experience into intervals (≤25, ≤50, ≤75, ≤100, ≤150, ≤200, ≤300) to explore trends in complication rates. All procedures were performed using the da Vinci Surgical System (Intuitive Surgical, Mountain View, CA), model Si. Eight different surgeons performed the hysterectomies, five of them treated women both with malignant and benign indications, while three operated only for benign indications. Each robotic surgeon performed approximately 20–40 RA‐TLHs annually. Antibiotic prophylaxis was administered to all women according to national guidelines. A conventional colpoprobe was used to mark the vaginal fornix until 2017, and later, a uterine manipulator (V‐Care (ConMed, Utica, NY, USA) or Advincula Delineator™ (CooperSurgical, Inc., Trumbull, CT, USA)) was used in most benign cases. Most women underwent surgery in a 30‐degree Trendelenburg position with an intra‐abdominal pressure of 12 mmHg. The colpotomy was performed with monopolar curved scissors using a maximum of 40 W cutting energy. The specimen was removed via the vagina if possible, alternatively morcellated vaginally, and in a few women, the ExCITE technique was used to remove the specimen. 12 The vaginal cuff was sutured in a single horizontal layer with barbed suture (V‐Loc™ 2–0, Medtronic, Minneapolis, MN, USA). Women were advised to abstain from sexual intercourse for 8 weeks post‐surgery. There was no routine follow‐up after surgery, but the women were strongly advised to contact the hospital in case of any postoperative adverse event. The normality of the data distribution was checked using the Kolmogorov–Smirnov test and Q‐Q plots, and non‐parametric tests were used as appropriate. The statistical analyses were performed using SPSS (Version 29.0.1; SPSS Inc., Chicago, IL, USA). We compared women with and without VCD using univariate and multivariate logistic regression analysis to identify risk factors for VCD. In the multivariable regression model, we included BMI, age, and uterine weight, which all were statistically significant factors for VCD in the univariable analysis. Duration of surgery, which also was significant using univariable analyses, was excluded due to its correlation with uterine weight (Pearson 0.366, p  < 0.001) and the need to adjust for multiple additional confounders (see Directed Acyclic Graph (DAG) analysis, Figure  S1 ). Additionally, including it did not improve model fit (Akaike Information Criterion (AIC)/Bayesian Information Criterion (BIC)). Statistical tests were two‐sided, and the significance level was set to 5%. Power analysis was not feasible.

Figure S1. Directed acyclic graph (DAG) analysis: Exposure: Length of surgery. Outcome: Vaginal cuff dehiscence. Length of surgery must be adjusted for age, BMI, surgeon’s experience, and malignancy in multivariable logistic regression.