The benefits of laparoscopic surgery are well known and include shorter hospital stay, less postoperative pain, quicker return to normal activities, fewer wound infections, and decreased blood loss in comparison with an abdominal approach.1 Given these clear benefits, the use of laparoscopy has increased substantially and is the preferred approach when vaginal hysterectomy is not feasible.2
Despite the increased utilization of laparoscopy for hysterectomy, conversion to laparotomy (or “conversion”) remains a risk and has been reported in 0–19% of patients.3 Reported risk factors for conversion include patient factors such as increasing age,3 increasing body mass index (BMI, calculated as weight (kg)/[height (m)]2),3–5 history of abdominopelvic surgery,4 presence of adhesions,4–6 endometriosis or leiomyomata,4 uterine weight3,5,6 as well as a less experienced surgeon.3 However, these risk factors are not consistently reported across all major studies,3–6 and prior studies have not examined the effect of the robotic surgical system. Furthermore, many of the prior studies are limited by small sample size and outcomes from a single institution.
The objective of this study was to evaluate the incidence and risk factors for conversion to laparotomy for both traditional laparoscopic and robotic hysterectomy performed for benign indications using a statewide multicenter prospective database. Our secondary objective was to determine differences in 30-day outcomes of women who had conversion to laparotomy. This information will enhance risk stratification and improve preoperative planning and patient selection for hysterectomy.
MATERIALS AND METHODS
All laparoscopic hysterectomies performed from January 1, 2013, through July 2, 2014, in the Michigan Surgical Quality Collaborative database were included in this analysis. The selected timeframe reflected the most complete data set at the time of the analysis. The Michigan Surgical Quality Collaborative consists of 52 academic and community hospitals voluntarily participating in this collaborative. Sixty percentage are community-based. Funded by Blue Cross and Blue Shield of Michigan/Blue Care Network, this represents approximately 30% of Michigan hospitals, and it includes patients from all insurance payers. At each participating hospital, a trained, dedicated nurse researcher reviews the entire medical record and uses a standardized data collection instrument to abstract patient characteristics, operative findings, complications, pathology results, and 30-day postoperative outcomes. Data were abstracted from clinic and hospital notes, operative reports, and pathology reports.
Cases collected in the Michigan Surgical Quality Collaborative database include general surgery procedures (such as appendectomy, cholecystectomy, colectomy) as well as hysterectomies. A standardized data collection methodology is used to reduce sampling bias. The year is divided into consecutive 8-day cycles so that each cycle starts on a different day. The intended effect is to promote surgeon sampling because it is common practice for surgeons to cluster procedures on 1 or 2 days of the week. At each hospital, the first 25 cases of a previously identified list of Current Procedural Terminology codes of an 8-day cycle are collected. The process of data abstraction and methodology is routinely validated through scheduled site visits, conference calls, and internal audits. The University of Michigan institutional review board provided “nonregulated” status to this study (HUM00073978).
All benign hysterectomies that were initiated with laparoscopy during the study period within the sample were included in this analysis. Laparoscopic hysterectomy included traditional laparoscopy, robotic-assisted (or “robotic”), and laparoscopic-assisted vaginal approaches. Patients with a known malignancy or obstetric indication were excluded from this analysis. Route of hysterectomy and conversion to laparotomy were obtained from operative reports and intraoperative records. The objective of this study was to evaluate the incidence and risk factors for conversion to laparotomy for both traditional laparoscopic and robotic-assisted laparoscopic hysterectomy performed for benign gynecologic indications.
We evaluated patient preoperative and perioperative characteristics. Patient demographics included age, parity, self-reported race (white or nonwhite), and type of insurance at the time of surgery. The patient's insurance was classified as private, Medicaid, Medicare, uninsured, or self-pay. Private insurance included Blue Cross Blue Shield of Michigan, Blue Care Network, health maintenance organization plans, and other private insurance plans. Medicare included Medicare only, Medicare with a supplemental plan such as Medigap insurance, or Medicare Advantage (Blue Cross Blue Shield or Blue Care Network of Michigan). Patients were considered to have Medicaid if they had Medicaid or a health maintenance organization Medicaid plan. Uninsured and self-pay were reported but not considered for additional multivariate analysis as a result of small numbers.
Additional patient characteristics included medical comorbidities (BMI, history of hypertension, deep vein thrombosis, preoperative blood transfusion), American Society of Anesthesiologists physical status classifications, surgical indications for hysterectomy, and history of prior abdominal or pelvic surgery. Surgical indications were determined based on the outpatient clinic notes, preoperative history, and physical or operative notes. These indications included abnormal uterine bleeding, leiomyomas, endometriosis, pelvic inflammatory disease, pelvic mass, chronic pelvic pain, and pelvic organ prolapse. Surgical indications were not mutually exclusive and patients could have more than one indication. Intraoperative information was based on operative notes and pathology notes. This included presence of abdominal or pelvic adhesions; presence of endometriosis anywhere in the pelvis; presence of endometriosis on the uterus, ovaries, or fallopian tubes; type of hysterectomy performed (total compared with supracervical); intraoperative complications; specimen weight; estimated blood loss in milliliters; and procedure duration (incision to closure) in minutes. The finding of unexpected malignancy, defined as hysterectomy performed for a benign indication and final pathology identified a malignancy, was also included. Adhesions were classified as none or mild, moderate, or severe. None or mild adhesions were those that were not mentioned in the operative report or described as “few” or “limited.” Moderate adhesions were described as “some,” “multiple,” or “many,” which may require lysis but do not impair the ability to do the operation. Severe adhesions were defined as “severe,” “dense,” “extensive,” “significant,” or “hostile”; taking an hour or more to lyse; or both. Severe adhesions were also defined as adhesions prohibiting the planned procedure.
Intraoperative complications included bowel, bladder, ureter, or vascular injury identified before completion of the hysterectomy. Specimen weight for the uterus was measured in grams and classified as less than 250, 250–499, 500–999 g, and greater than or equal to 1,000 g. Estimated blood loss was classified as less than or equal to 100, 101–300, 301–500, 501–1,000 mL, and greater than 1,000 mL. The duration of surgery was split into 2-hour increments ranging from less than 2 hours to greater than 6 hours.
Hospital characteristics included teaching status, as defined by the 2012 American Hospital Association survey. Hospital bed size was defined as small (less than 300 beds), medium (300–499 beds), and large (500 beds or greater).
Surgeon volume was determined by the number of all hysterectomies contributed to the Michigan Surgical Quality Collaborative sample in the prior 24 months by each surgeon. This included abdominal, laparoscopic, and vaginal hysterectomies. This was used as a proxy for total surgeon volume because the Michigan Surgical Quality Collaborative does not capture every surgery performed at each hospital, only the first 25 cases of each 8-day data-capture cycle. Thus, a surgeon's total annual caseload was not available and we considered this measure the best available proxy. We then examined the distribution of surgeon volume and divided the surgeons into equal volume-based tertiles of low, medium, and high volume, as has previously been reported.7–9 The distribution of cases in each tertile can be seen in Table 1. The lowest volume tertile contributed one to four hysterectomies in the 24-month period, the intermediate-volume tertile contributed 5–17, and the high-volume tertile contributed 18 or more hysterectomies to the sample in the prior 24 months. Given that the top tertile still included a low number of hysterectomies contributed to the sample, we also examined ultrahigh-volume surgeons by dividing the surgeons into those below the 75th percentile, 75th–89th percentile, 90th–94th percentile, 95th–98th percentile, and 99th percentile or greater.
Our secondary objective was to determine differences in 30-day outcomes of women who had conversion to laparotomy, including the incidence of surgical site infection, postoperative transfusion, venous thromboembolism, readmission, and reoperation. This information was abstracted from hospital notes, office notes, laboratory results, radiology results, nursing notes, and emergency department documentation. Surgical site infection definitions were based on definitions from the Centers for Disease Control and Prevention surveillance definition of health care-associated infections.10 Sepsis was defined as a recent history of new infection within 30 days postoperatively with any two of the following signs and symptoms: temperature greater than 38.3°C or less than 36°C, heart rate greater than 90 beats per minute, respiratory rate greater than 20 breaths per minute, white blood cell count greater than 12,000 cells per cubic millimeter or less than 4,000 cells per cubic millimeter, hyperglycemia (plasma glucose greater than 120 mg/dL) in the absence of diabetes, or acutely altered mental status.11
Distributions for continuous variables were checked for normality looking at skew and kurtosis. Continuous variables were examined and extreme outliers were identified and removed. Approximately normally distributed data were reported as mean±standard deviation and nonparametric data were reported as median (interquartile range). Wilson-binomial 95% confidence intervals were calculated for proportions where appropriate. Descriptive analysis of categorical variables was conducted using χ2 or Fisher exact test in the case of small cell sizes and Student t test or analysis of variance for continuous variables as appropriate.
Clinically relevant covariates were considered for multivariable logistic regression to examine likelihood of conversion from laparoscopy to laparotomy. Candidate covariates were assessed for missing data and those with sufficient data for model development were considered for analysis. Although parity and history of prior abdominal surgeries were clinically relevant covariates, they were excluded as a result of a significantly larger proportion of cases missing compared with other covariates considered for bivariate and multivariate analyses. To account for clustering of patients within hospitals, Huber-White robust standard errors were calculated for each of the parameter estimates. For categorical variables with more than two categories, dummy variables were derived to measure direct effect sizes. Because collinearity resulting from a significant correlation can dramatically affect parameter estimates and effect sizes for the model, Spearman or Pearson correlation matrices for all variables considered in a model were obtained. Iterative variable selection taking into account collinearity and clinical relevance of selected variables led to a reduced model with a strong C-statistic (concordance). Final model diagnostics included decile and quintile analysis comparing observed and adjusted rates for conversion. Logistic regression models were used to calculate the predicted incidence of conversion among robotic and traditional laparoscopic cases adjusted for the variables included in the final model. A secondary analysis was performed that fit the previously defined model in addition to an interaction term. This interaction term accounted for the relationship between surgical approach (robotic-assisted laparoscopy compared with traditional laparoscopy) and surgeon volume. This was used to calculate the predicted risk of conversion based on surgical approach across surgeon volume groups. Data analyses were performed using Stata 14.0.
A total of 12,122 hysterectomies were available in the data set from January 1, 2013, to July 2, 2014. After exclusion of cases for surgical approach (2,832 abdominal, 1,343 vaginal, and one not discernible as a result of a lack of data), an absence of pathology data (232), gynecologic cancer (719), and obstetric indications (3), there were 6,992 hysterectomies eligible for inclusion (Fig. 1). Of these, 6,717 were completed laparoscopically, and 275 were converted to an open procedure, for an overall rate of conversion of 3.93%. Seven of these cases were initiated emergently, and none of those were converted to laparotomy.
Univariate analyses are shown in Tables 2 and 3. Multiple preoperative characteristics (Table 2) were associated with a significantly increased risk of conversion, including BMI greater than 30 and preoperative indication of leiomyomas, pelvic inflammatory disease, or a pelvic mass. Factors that were associated with decreased odds of conversion included previous failure of an alternative treatment and preoperative indication of pelvic organ prolapse. Compared with moderate-volume surgeons, high-volume surgeons had lower odds of conversion in the univariate analysis. In addition, the rate of conversion was significantly higher in traditional laparoscopy than robotic-assisted laparoscopy (8.28% compared with 1.57%, P<.001). Those who were self-pay or uninsured had an increased odds of conversion compared with those with private insurance (odds ratio [OR] 2.98, 95% confidence interval [CI] 1.47–6.04). This was not included in the multivariate analysis as a result of small numbers. Compared with those who underwent traditional laparoscopy, the group of patients undergoing robotic-assisted laparoscopic hysterectomy had characteristics associated with higher surgical complexity, with statistically significantly higher BMIs and more frequent removal of the cervix and presence of endometriosis. There was no difference on other covariates including specimen weight and adhesion severity.
As shown in Table 3, intraoperative factors that were associated with an increased odds of conversion included presence of either moderate or severe adhesions or unexpected malignancy. Presence of endometriosis on the uterus, ovaries, or fallopian tubes also increased the risk of conversion. The risk of conversion was significantly higher with any complication including bowel, bladder, ureteral, or vascular injury. Increasing specimen weight, blood loss, and operative time all were associated with significantly increased odds of conversion.
Patients who underwent concurrent procedures such as oophorectomy or hernia repair were not at increased odds of conversion (Table 3). Having concurrent bowel surgery (n=5) was associated with conversion in the bivariate analysis. However, on further investigation of these cases, all of these bowel surgeries appeared to be initiated as a result of an intraoperative bowel complication. This was reflected in the documentation of bowel injury by Current Procedural Terminology codes. Therefore, this was not included in the model.
As shown in Table 4, the significant risk factors for conversion with multivariate logistic regression modeling were age older than 40 years and 60 years or younger, BMI greater than or equal to 30, preoperative indications of pelvic mass, presence of moderate or severe adhesions, and specimen weight greater than 250 g. The factors most strongly associated with decreased odds of conversion in the multivariate model were having a robotic procedure and having a high-volume surgeon. Other factors that decreased the risk of conversion included having an alternative treatment before hysterectomy or a preoperative indication of pelvic organ prolapse. This model was then used to estimate predicted means of conversion based on route of surgery and surgeon volume. After arriving at a parsimonious model, the C-statistic was 0.85 and the Hosmer-Lemeshow test statistic was 8.60 with a P value of .38 for 10 groups. Using this model, the predicted risk of conversion to laparotomy with traditional laparoscopy compared with robotic-assisted laparoscopy was 5.4% compared with 0.8% (P<.001) after adjusting for all other variables in the model. High-volume surgeons were less likely to convert to laparotomy compared with low- and medium-volume surgeons with a predicted risk of conversion of 1.4% compared with 2.25% (P=.015).
The high-volume tertile surgeons performed 18–258 hysterectomies captured within the 24-month sample. We identified a significant correlation between use of robotic surgery and surgical volume with a significantly greater proportion of high-volume surgeons using the robotic platform (72.02%) compared with the low-volume surgeons (49.09%, P<.001). Given some collinearity between the use of the robotic surgical platform and high-volume surgeons, we then performed a subanalysis of high-volume surgeons to examine the effect of robotic use in high-volume surgeons. Even among high-volume surgeons, the odds of conversion was lower with the robotic procedure (7.54% compared with 1.46%, P<.001; adjusted OR 0.13, 95% CI 0.06–0.27), even when controlling for other factors including uterine weight and adhesive disease.
To further examine the relationship among high-surgeon volume, use of robotic surgery, and the effect on conversion, a secondary analysis was conducted among ultrahigh-volume surgeons. We divided surgeon volume into those below the 75th percentile, 75th–89th percentile, 90th–94th percentile, 95th–98th percentile, and 99th percentile or greater. The number of hysterectomies contributed to the sample in 24 months in each group was 1–23, 24–41, 42–60, 61–127, and 128–258, respectively. After adjusting for patient risk factors, surgical approach (robotic-assisted laparoscopy compared with traditional laparoscopy), surgeon volume categories, and significant interaction between approach and volume, we calculated the adjusted predicted risk of conversion with the robotic surgical system compared with traditional laparoscopy (Fig. 2). This demonstrated that there remained a benefit of use of robotics even among ultrahigh-volume surgeons.
Complications after conversion were examined (Table 5), and those who had a conversion had increased risk of incisional infections, postoperative transfusion, postoperative severe sepsis (sepsis with organ dysfunction), and need for reoperation within 30 days. There was no difference in rates of postoperative organ space surgical site infection, pulmonary embolism or deep vein thrombosis, sepsis, or readmission within 30 days. We found that intraoperative complications, which may have influenced the decision to convert to laparotomy, occurred in less than 20% of those who had a postoperative complication. Therefore, the association between conversion and postoperative complications was not entirely related to the prior occurrence of an intraoperative complication.
In this regional collaborative of 52 hospitals, there was a sevenfold reduction in the odds of conversion to laparotomy with use of robotic-assisted laparoscopy compared with traditional laparoscopy. The avoidance of conversion in our cohort of patients had important clinical repercussions. Patients who had conversion were more likely to experience surgical site infection, blood transfusion, severe sepsis, and reoperation even when no prior intraoperative complication occurred. Similar results were seen in the colorectal surgery literature with poorer outcomes in morbidity, mortality, blood transfusion, and postoperative hospital stay.12,13
A wide range of conversion rates for hysterectomy has been reported, ranging from 0% to 19%.3 The 2014 Cochrane Review, which pooled the outcomes from four randomized controlled studies, did not find a difference in conversion rates of robotic and traditional laparoscopies (3.55% compared with 2.98%). That analysis involved 337 patients with 11 conversions.15 This small sample size may have insufficient power to detect a difference between the approaches. Furthermore, this analysis only included surgeries performed by 13 high-volume surgeons from four tertiary care centers. In contrast, our analysis involved 6,992 hysterectomies from a sample of 638 surgeons at community and tertiary academic hospitals with bed size ranging from less than 100 to greater than 1,000. The larger, diverse sample size in our study may explain why we found a difference where authors of the Cochrane Review did not. This lower conversion rate using the robotic surgical system has also been seen in other surgical specialties with decreased rates of conversion using robotics for prostatectomy and surgery for colorectal cancer.13,15–17 Consistent with prior studies, surgeon volume was found to be associated with lower odds of conversion,3,18,19 and increasing BMI, adhesive disease, and increasing uterine weight were all found to be associated with increased odds of conversion.3,4,6,20
Strengths of this analysis are a large sample of hysterectomies from a statewide database that includes all payer groups, academic, and community hospitals. However, we do not know the indication for conversion and cannot differentiate between conversions as a result of an adverse, emergent event and those without complication and related to surgeon judgment.21 The indication for conversion has been associated with different risks of complications and length of hospital stay.3 There is also an inherent limitation of the sampling methodology, which captures a random sample of patients at each institution and not every patient for each surgeon. Although our surgeon volume variable likely reflects the relative range of surgical experience, surgeon skill and decision-making cannot be ascertained from a surgical database, and this analysis is limited to the available variables and cases included in the Michigan Surgical Quality Collaborative database. We were also unable to determine the patient distribution of surgical approach by any given surgeon. Also, many complications are rare events after hysterectomy and even larger samples may be necessary to detect a difference in rare outcomes such as venous thromboembolism. Lastly, the population and practice patterns in Michigan may not be applicable to other geographic regions.
In summary, our study showed that more than 96% of all hysterectomies for benign indications initiated laparoscopically are completed laparoscopically. Use of the robotic platform and higher surgeon volume were both significantly and independently associated with decreased odds of conversion. Furthermore, conversion to laparotomy is associated with increased risk of morbidity. Although previous studies22 including the Cochrane Review have demonstrated no benefit and higher cost to robotic surgery compared with traditional laparoscopy, many studies were limited by smaller samples and single high-volume institutions and prior cost data do not account for the potential increased morbidity and cost associated with conversion.23 Indeed, this study demonstrates a potential advantage of robotic surgery in a generalizable population. Although the relationship between the use of the robotic surgical platform and surgeon volume is complex and interrelated, there appears to be a sustained lower predicted risk of conversion in robotic hysterectomy across surgeon volumes, including high-volume surgeons. Future research should further examine the association among surgical volume, surgical approach, morbidity, and cost associated with conversion to laparotomy.
Although the findings of this study suggest a significant relationship among the robotic platform, high-volume surgeons, and lower odds of conversion to laparotomy, these results are not intended to define medical policies or surgical privileging. Indeed, the balance between high-quality surgical care and access to quality care is complex. Our specialty needs to further address the effect of decreased surgical training, increased skill required to offer minimally invasive options to patients with complex pelvic pathology, and clinical practices with low surgical volume on quality of care.
1. Nieboer TE, Johnson N, Lethaby A, Tavender E, Curr E, Garry R, et al. Surgical approach to hysterectomy for benign gynaecological disease. The Cochrane Database of Systematic Reviews 2009, Issue 3. Art. No.: CD003677. DOI: 10.1002/14651858.CD003677.pub4.
2. Choosing the route of hysterectomy for benign disease. ACOG Committee Opinion No. 444. American College of Obstetricians and Gynecologists. Obstet Gynecol 2009;114:1156–8.
3. Twijnstra AR, Blikkendaal MD, van Zwet EW, Jansen FW. Clinical relevance of conversion rate and its evaluation in laparoscopic hysterectomy. J Minim Invasive Gynecol 2013;20:64–72.
4. Leonard F, Chopin N, Borghese B, Fotso A, Foulot H, Coste J, et al. Total laparoscopic hysterectomy: preoperative risk factors for conversion to laparotomy. J Minim Invasive Gynecol 2005;12:312–7.
5. Song T, Kim TJ, Kang H, Lee YY, Choi CH, Lee JW, et al. Factors associated with complications and conversion to laparotomy in women undergoing laparoscopically assisted vaginal hysterectomy. Acta Obstet Gynecol Scand 2012;91:620–4.
6. Park SH, Cho HY, Kim HB. Factors determining conversion to laparotomy in patients undergoing total laparoscopic hysterectomy. Gynecol Obstet Invest 2011;71:193–7.
7. Wallenstein MR, Ananth CV, Kim JH, Burke WM, Hershman DL, Lewin SN, et al. Effect of surgical volume on outcomes for laparoscopic hysterectomy for benign indications. Obstet Gynecol 2012;119:709–16.
8. Wright JD, Lewin SN, Deutsch I, Burke WM, Sun X, Herzog TJ. Effect of surgical volume on morbidity and mortality of abdominal hysterectomy for endometrial cancer. Obstet Gynecol 2011;117:1051–9.
9. Rogo-Gupta LJ, Lewin SN, Kim JH, Burke WM, Sun X, Herzog TJ, et al. The effect of surgeon volume on outcomes and resource use for vaginal hysterectomy. Obstet Gynecol 2010;116:1341–7.
10. Horan TC, Andrus M, Dudeck MA. CDC/NHSN surveillance definition of health care-associated infection and criteria for specific types of infections in the acute care setting. Am J Infect Control 2008;36:309–32.
11. Dellinger RP, Levy MM, Rhodes A, Annane D, Gerlach H, Opal SM, et al. Surviving Sepsis Campaign: international guidelines for management of severe sepsis and septic shock. Intensive Care Med 2013;39:165–228.
12. Marusch F, Gastinger I, Schneider C, Scheidbach H, Konradt J, Bruch HP, et al. Importance of conversion for results obtained with laparoscopic colorectal surgery. Dis Colon Rectum 2001;44:207–14.
13. Tanis PJ, Buskens CJ, Bemelman WA. Laparoscopy for colorectal cancer. Best Pract Res Clin Gastroenterol 2014;28:29–39.
14. Liu H, Lawrie TA, Lu D, Song H, Wang L, Shi G. Robot-assisted surgery in gynaecology. The Cochrane Database of Systematic Reviews 2014, Issue 12. Art. No.: CD011422. DOI: 10.1002/14651858.CD011422.
15. Finkelstein J, Eckersberger E, Sadri H, Taneja SS, Lepor H, Djavan B. Open versus laparoscopic versus robot-assisted laparoscopic prostatectomy: the European and US experience. Rev Urol 2010;12:35–43.
16. Trastulli S, Farinella E, Cirocchi R, Cavaliere D, Avenia N, Sciannameo F, et al. Robotic resection compared with laparoscopic rectal resection for cancer: systematic review and meta-analysis of short-term outcome. Colorectal Dis 2012;14:e134–56.
17. Ahlering TE. Robotic versus laparoscopic radical prostatectomy. Nat Clin Pract Urol 2004;1:58–9.
18. Doll KM, Milad MP, Gossett DR. Surgeon volume and outcomes in benign hysterectomy. J Minim Invasive Gynecol 2013;20:554–61.
19. Twijnstra AR, Blikkendaal MD, van Zwet EW, van Kesteren PJ, de Kroon CD, Jansen FW. Predictors of successful surgical outcome in laparoscopic hysterectomy. Obstet Gynecol 2012;119:700–8.
20. Sokol AI, Chuang K, Milad MP. Risk factors for conversion to laparotomy during gynecologic laparoscopy. J Am Assoc Gynecol Laparosc 2003;10:469–73.
21. Blikkendaal MD, Twijnstra AR, Stiggelbout AM, Beerlage HP, Bemelman WA, Jansen FW. Achieving consensus on the definition of conversion to laparotomy: a Delphi study among general surgeons, gynecologists, and urologists. Surg Endosc 2013;27:4631–9.
22. Wright JD, Ananth CV, Lewin SN, Burke WM, Lu YS, Neugut AI, et al. Robotically assisted vs laparoscopic hysterectomy among women with benign gynecologic disease. JAMA 2013;309:689–98.
23. Healy MA, Mullard AJ, Campbell DA Jr, Dimick JB. Hospital and payer costs associated with surgical complications. JAMA Surg 2016 May 11 [Epub ahead of print].