Effects of Obesity on Peri- and Postoperative Outcomes in Patients Undergoing Robotic versus Conventional Hysterectomy

S

The use of robot-assisted laparoscopic surgery has gained widespread application in the field of gynecology despite the lack of evidence that it improves outcomes [1]. Similar to urologic and colorectal surgery, robot-assisted laparoscopy in gynecology has primarily been used for malignant disease. The role of robot-assisted hysterectomy (RTLH) for benign disorders is limited [2]. In a Cochrane review on RTLH, comparing it with traditional laparoscopic hysterectomy (TLH), a mere 550 patients were included in a comparison of postoperative complications despite the 300 000 or more gynecologic robot-assisted procedures that have been performed worldwide [1,3]. In hysterectomies in general, previous studies have shown that the complication rate is highest for open abdominal surgery but does not differ significantly among the minimally invasive modalities. Vaginal hysterectomy (VH) has the shortest duration of the procedures and is recommended as first choice, when applicable [4−8].
Hysterectomy in women with obesity is considered more complex owing to difficulties in finding anatomic landmarks, a deeper surgical field, and a narrower pelvis. In accordance with this perception, studies have shown that women with obesity undergoing hysterectomy are at increased risk of complications compared with women with normal weight [9−11]. Minimally invasive hysterectomy seems to have a lower rate of infections and estimated blood loss, than open abdominal hysterectomy (AH) [5−7]. However, it is largely unknown how robot-assisted hysterectomy influences the risk of complications among women with obesity. The aim of the present study was to assess if women with obesity have increased morbidity and complication rates compared with women with normal weight undergoing hysterectomy for benign indications and if the hysterectomy mode affects the outcomes.

Materials and Methods
Data were collected and linked from 3 Swedish population-based registers: the National Quality Registry for Gynecological Surgery, the National Patient Register, and the Prescribed Drug Register. The National Quality Registry for Gynecological Surgery [12] is a register with a coverage of 88.9% to 89.8% of all hysterectomies performed in Sweden, reported by 20 to 21 of 21 (95%−100%) counties, during the study period. It contains information of demographics; patient-reported outcomes; and clinical data before, after, and during surgery. The surgeon reports the clinical data before and directly after the operation. When the patient leaves the hospital, a second data form is filled in by the gynecologist regarding complications during the hospital stay. The diagnosis is filled in when the histopathologic result is received. If the patient responds to the questionnaire, from 2 months to 1 year after the surgery, stating that she had a complication, a notification is sent to the surgeon or, in some hospitals, to a coordinating gynecologist. This person will read the patient's medical record and then register the complication in a standardized way (see Appendix for a detailed description), if judged relevant. In this way the register will contain both patient-and surgeonreported complications.
Data regarding complications after surgery are sometimes underreported. Therefore, we added data from 2 other national registers, supervised by the Swedish Board of Health and Welfare, to make sure that the data regarding complications would be as complete as possible. The National Patient register achieved nationwide coverage in 1987. The register records all in-patient diagnoses, which are based on the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision, as well as up to 12 operation codes according to the Swedish Classification of Operations and Major Procedures. Ninetyeight percent of records achieved the correct coding for surgical procedures with less than 1% yearly loss to registration when cross-checked with hospital charts [13].
The Prescribed Drug Register [14] contains information about prescribed and retrieved drugs from pharmacies. The register has a coverage of between 97% and 100%, depending on variables.
Women undergoing a total hysterectomy for benign reasons with or without a salpingo-oophorectomy between January 1, 2015, and December 31, 2017, were identified from the surgical register. Women who had had a hysterectomy for obstetric or malignant indication for hysterectomy as well as pelvic organ prolapse were excluded. Using the individually unique national registration number assigned to all Swedish nationals, every woman was subsequently linked to the National Patient and Prescribed Drug registers for data extraction.
The primary outcome was the occurrence of peri-and postoperative complications, including reoperations and readmissions, in relation to the body mass index (BMI) class 1 year after the hysterectomy. Second, we evaluated how the BMI affected operation time, blood loss, blood transfusions, rate of conversion to laparotomy, length of hospital stay, amount of pre-and postoperative use of analgesics, and number of days to normal activities of daily living.
The information regarding BMI derived from the surgical register and patients was grouped according to the World Health Organization's classification of obesity: underweight (BMI <18.5); normal weight (BMI 18.5 −24.9); preobesity (BMI 25−29.9); obesity class I (BMI 30.0−34.9); obesity class II (BMI 35.0−39.9); and obesity class III (BMI ≥40). Women with obesity class II and III were merged into 1 group. When obesity class is not specified in the article, the words "obesity" and "obese" are defined as BMI ≥30. To investigate changes in the prescribed drug, prescription data were collected 3 months before and 4 months after surgery. Different drugs were identified by the Anatomical Therapeutic Chemical Classification code. Data from the National Patient register were collected for up to 1 year after surgery to be able to identify readmissions for up to 4 months and reoperations for up to 1 year after surgery (i.e., until December 31, 2018).
Hysterectomy was classified according to the mode of surgery as per the Swedish Classification of Operations and Major Procedures into open AH, traditional laparoscopic hysterectomy (TLH), VH, and robot-assisted laparoscopic hysterectomy (RTLH). TLH was defined as all laparoscopic hysterectomies and included total laproscopic hysterectomy, laparoscopic hysterectomy with suturing of the vagina transvaginally, and laparoscopically assisted VH.
Peri-and postoperative complications associated with hysterectomy were identified from the surgery register and grouped into thrombosis, postoperative infections (divided into wound infections and vaginal infections), postoperative bleeding complications, and intra-abdominal organ injury (with a subanalysis of injury to the urinary bladder, ureter, or intestine). Data ascertainment on complications and adverse events during 1-year follow-up after hospital release were derived from several sources and are described in detail in the Appendix.
The baseline demographic and descriptive data included age, American Society of Anesthesiologists (ASA) Physical Status classification (grouped as ASA class 1−2 or 3−5), parity (grouped as 0, 1−2, and ≥3 children), previous cesarean section, smoking habits, previous abdominal surgery, method of anesthesia, antibiotic and thromboprophylaxis, concurrent salpingo-oophorectomy, concurrent procedures at the time of the hysterectomy, uterus size (in grams), hospital procedure volume per year (grouped as <50, 50−150, and >150), and surgeon procedure volume per year (grouped as <10, 10−20, 21−50, and >50). A subgroup analysis of the effects of various modes of hysterectomy in relation to the BMI class was performed using the definitions of surgery mentioned above.
We used Stata software version 15.0 (StataCorp LLC, College Station, TX) to perform the statistical analyses. Demographic data and perioperative descriptive data were presented as frequencies and proportions for categoric variables. For the categoric end point we used both univariate and multivariate regression models, including baseline demographic data. To account for intraindividual dependence, the robust sandwich estimator for standard error was used. Multivariate regression was conducted as a stepwise procedure in which variables from the univariate model with a p-value >.25 were excluded from the multivariate regression model. The results from the univariate regression model are presented as crude odds ratio (cOR) and from the multivariate regression model as adjusted odds ratio (aOR), with a 95% confidence interval (CI). All pvalues <.05 were considered significant.
The BMI data were missing in 34.5% of the cases. Because we had a high frequency of missing data in our primary predictor, BMI, we performed multiple imputation by chained equations (MICE) to study this further, in addition to conducting a complete data analysis [15]. MICE assumes that the missing information for any given variable used in the imputation model is missing at random. In particular, we also performed MICE generating 20 complete artificial datasets. The results from these 20 complete data analyses were combined to give an outcome of a single imputation model. Consequently, our imputation model results were based on fully observed variables as well as variables with imputed values where the missing information had occurred. We included all variables from Tables 1 and 2.
The study was approved by the research ethics committee at Karolinska Institutet, Stockholm, Sweden and conforms to the STROBE guidelines for reporting observational studies (www.strobe-statement.org).
The frequency of complications with regard to the BMI class and to the mode of hysterectomy in women with obesity are presented in Tables 2A and 2B, respectively. Within 1 year after hysterectomy, the overall frequency of complications was higher in women with obesity class II−III (aOR 1.4; 95% CI, 1.1-1.8) than those in women with normal weight. In women with obesity, infection was the most common complication, in particular, wound infections (Table 2A). Overall complications 1 year after hysterectomy in women with obesity class I−III occurred most frequently in AH (27.2 %), thereafter in TLH (22.4 %), VH (17.9 %), and RTLH (16.9 %). The difference in complication rates between AH and RTLH was significant (aOR 1.8; 95% CI, 1.2−2.6) in women with obesity. Furthermore, intraoperative complications were more common in the VH group versus the RTLH group in women with obesity class I−III (aOR 4.4;95% CI, 1.2−15.8). The most common intraoperative complication in women with obesity undergoing VH was injury to the urinary bladder (5 out of 9). There were no other significant differences with regard to the frequency of intraoperative and overall complications comparing the different modes of hysterectomy.
Tables 3A and 3B show intraoperative data with regard to the BMI class and to the mode of hysterectomy in women with obesity. Women with obesity class II−III had a higher risk of conversion to AH (aOR 2.2; 95% CI, 1.3−3.7) than women with normal weight. When comparing the hysterectomy modes, women with obesity class I−III had 28-fold higher risk of conversion with TLH and 17-fold higher risk of conversion with VH (aOR 28.2; 95% CI, 6.4−124.7 and aOR 17.1; 95% CI, 3.5−83.8, respectively) than that with RTLH. In most of the cases (71/80), the reason for conversion stated in the register was "technical difficulties." Descriptive data on converted cases can be found in the Appendix (Supplementary Table 1).
The higher the BMI class, the higher the proportion of women bleeding >500 mL and an operation time >2 hours. In women with obesity class I−III, AH, TLH, and VH were associated with a higher risk of estimated blood loss >500 mL than RTLH (aOR 11.8; 95% CI, 3.4−40.5; aOR 8.5; 95% CI, 2.5−29.5; and aOR 5.8; 95% CI, 1.5−22.8, respectively). In obesity class I−III, the frequency of procedures with a duration of hysterectomy more than 2 hours was more than 4-fold higher with TLH than with RTLH (aOR 14.5; 95% CI, 4.2−50.0). In contrast, VH was associated with a lower proportion of women with obesity class I −III undergoing surgery >2 hours compared with RTLH (aOR 0.1; 95% CI, 0.1−0.3) ( Table 3B).
The postoperative outcomes are presented in Tables 4A  and 4B. Women with obesity undergoing AH or TLH more often had a hospital stay >2 days, and women in the AH group more often required 3 to 9 days to return to normal activities of daily living and had more frequently prolonged pain than women in the RTLH group. There was no difference in the frequency of reoperations or readmission with regard to the BMI class or hysterectomy mode in patients with obesity. RTLH in women with obesity was performed by surgeons with procedure volume of less than 20/year in 157 (63%) cases versus 236 (74 %) cases of TLH.

Discussion
The main findings of this study were that in women undergoing hysterectomy, obesity is associated with a higher overall rate of complications, perioperative bleeding, and conversion to abdominal surgery. Furthermore, in women with obesity, RTLH is associated with less perioperative bleeding, as well as a many-fold lower rate of conversion to open surgery, than conventional hysterectomy. In women with obesity, AH was associated with a higher overall complication rate, and VH had a slightly higher risk of intraoperative complications, both in comparison with RTLH [16].
There is a range of studies comparing different surgical modes and routes of hysterectomy for benign indication [1,17−19], but only a few focusing on the impact of obesity Flowchart of the study population.

Brunes et al.
Effects of Obesity on Peri-and Postoperative Outcomes [10,11,20,21]. In agreement with our findings, previous studies found a higher rate of postoperative morbidity in patients with obesity undergoing hysterectomy [21], of which the most commonly reported complication is postoperative infections [22]. In a cohort study by Lim et al [16] evaluating the risk factors for conversion to AH in comparison with TLH and RTLH, the investigator concluded that obesity was a risk factor for conversion to AH and that TLH had a higher conversion rate than RTLH. This concurs with our results showing that the conversion rate was many times higher in both VH and TLH than in RTLH in women with obesity. Our study also shows a lower estimated blood loss in favor of the robot-assisted method compared with other minimally invasive methods, which differs from  Data are presented as mean ( § standard deviation), median (interquartile range) or frequencies (%). * Other concurrent surgery, not including surgery on uterus or adnexa, for example, suture of the ureter, enterorrhaphy, or colostomy.

Table 2B
Complications   previous studies with uncomplicated benign hysterectomy [19]. A reason for this discrepancy could be that the previous randomized controlled studies were underpowered to analyze the estimated blood loss in different BMI classes. In addition, in uncomplicated hysterectomies, the operation method probably is of less importance with regard to blood loss than in more complex cases. Robotic surgery has often been criticized for a long operation time when compared with other methods of hysterectomy. Our data support this assumption with regard to VH, but women with obesity undergoing TLH actually had a longer operation time than women undergoing RTLH. In our data, there was a higher rate of low-volume surgeons in the TLH group than in the RTLH group. In the analysis, we have adjusted for surgeon volume as well as hospital volume. Nevertheless, the differences in operation time between TLH and RTLH might partly be due to the high number of hysterectomies performed by low-volume surgeons in this cohort. One could speculate that either a faster learning curve or the additional training program before being introduced to robot-assisted surgery could be an explanation for these findings. There is also a possibility that robot-assisted surgeons are more experienced in general. In contrast, in our experience in Sweden, RTLH is more often chosen in complex cases owing to the higher costs of this hysterectomy route.
Surgery in women with obesity is challenging [20], and women with obesity have a higher risk of complications,  [10], a decision often taken on the basis of the skill sets and technical resources available at the local hospital rather than the optimal choice for the individual patient on the basis of her defining characteristics. In women with obesity, there seem to be significant advantages with robotic surgery compared with other methods of hysterectomy, which should be taken into consideration when planning for surgery.
In the future, a randomized controlled multicenter study comparing minimally invasive methods and, in particular, TLH with RTLH in women with obesity is warranted. Such studies should also include the cost-benefit and cost-effectiveness aspects of the procedures, given the high costs of acquiring robot technology.
The prospective data ascertainment, a large-scale population-based design, and the uniform classification of both exposure and outcome are some of the strengths of our study. Recall bias was prevented by the independent prospective data ascertainment, and the near-complete registration of available patients during the observational period reduced the risk of reporting and ascertainment bias. We also consider the risk of misclassification of exposure and outcome to be limited, given the high quality of the registers in use. Even so, we recognize that the procedures are not entirely standardized and may include some diversity. This may be especially true for laparoscopic hysterectomies, which, in some clinics, involve a combination of laparoscopic and vaginal approaches. We did not have information on some potential effect modifiers or confounders, including alcohol consumption. ASA class may not be sufficient as an estimate for comorbidity. In addition, a limitation of this study was the relatively high rate of missing data on individual BMI. We could not identify any systematic bias underlying the missing BMI values, and we were also unable to detect any systematic differences between patients who reported BMI and those who did not. When we used multiple imputation, the results were largely consistent with the complete analysis, suggesting that the missing BMI data did not affect our estimates. This implies that the propensity for missing data is not related to the missing data itself (missing at random) and was unrelated to its hypothetical value or the value of other variables (missing completely at random). However, we cannot entirely rule out that the missing data are not missing at random and acknowledge that the results from multiple imputation should be interpreted with caution (Supplemental  Tables 2A and 2B).

Conclusion
The use of RTLH may lower the risk of conversion rates and intraoperative bleeding in women with obesity compared with other modes of hysterectomy. The complication rates were the highest in women with obesity undergoing AH.

Supplementary materials
Supplementary material associated with this article can be found in the online version at https://doi.org/10.1016/j. jmig.2020.04.038.