Abstract
Background:
Robotic-assisted proctectomy (RAP) has been reportedly associated with lower rates of conversion to laparotomy than laparoscopy in several cohort studies. This st0udy aimed to assess the temporal trends in conversion from RAP to laparotomy stratified by patient and treatment-related factors.
Methods:
This retrospective observational study was undertaken to analyse the temporal trends in unplanned conversion from RAP to laparotomy. Changes in the rates of conversion over time were plotted as line graphs, and the significance of each trend was calculated with the Cochran-Armitage trend test. A case-control analysis of factors associated with conversion to open surgery was conducted.
Results:
The study included 23,644 patients (62.3% male, median age: 60 years). 1280 (5.4%) patients were converted to laparotomy. There was a significant linear trend of decreased conversion over time (3.9% in 2021 compared with 10.4% in 2010; P < .001). The reduction in conversion rates was significant in all patients except in patients <50 years (P = .838), Black patients (P = .358), patients with a Charlson comorbidity index score >1 (P = .053), patients with governmental insurance other than Medicaid and Medicare (P = .629), and patients undergoing abdominoperineal resection (APR) (P = .129) or pelvic exenteration (PE) (P = .326). The independent predictors for increased conversion were male sex, higher Charlson scores, community cancer programs, comprehensive community cancer programs, household income of <$63,000, tumors ≥5 cm, and PE.
Conclusions:
Unplanned conversion from RAP to laparotomy showed a linear trend of reduction over time, which was statistically significant except in young patients, Black patients, patients with significant comorbidities, and patients undergoing APR or PE.
Introduction
Minimally invasive surgery, including laparoscopic and robotic-assisted surgery (RAS), has become the standard of care for colorectal cancer (CRC) in numerous hospitals worldwide.1,2 The use of RAS in several surgical disciplines has significantly increased from 1.8% in 2012 to 15.1% in 2018. 3 The increased use of RAS was also noted in colorectal surgery, as recent national database studies found a significant increase in robotic-assisted proctectomy (RAP) over time from 5.2% to 28.4% and robotic-assisted colectomy from 2.9% to 12.7%.4,5 RAS has been suggested to reduce the rates of conversion to open surgery. Although cohort studies demonstrated a lower likelihood of conversion to laparotomy with robotic-assisted colorectal surgery than laparoscopy;6–8 the Robotic versus Laparoscopic Resection for Rectal Cancer (ROLARR) trial, which was primarily designed to assess conversion rates, did not confirm the same. 9
Unplanned conversion from minimally invasive to open surgery for CRC may have an adverse impact on short-term and survival outcomes. Our previous study 10 demonstrated that conversion to open surgery was associated with higher 30- and 90-day mortality and unplanned readmission, longer hospital stays, and reduced survival. Therefore, it is important to assess the risk factors for and the trends in unplanned conversion in colorectal surgery. An American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) database analysis found that male patients aged ≥65 years with obesity and poor functional status undergoing emergency surgery were more likely to be converted from RAS to laparotomy. 11 Another study 12 noted that the risk factors for conversion may differ between laparoscopic and robotic-assisted colorectal surgery; however, in both groups, a higher surgeon volume was less likely associated with conversion.
In addition to understanding the risk factors for conversion, it is important to understand the trends in conversion over time. While conversion rates are expected to decrease over time because of increased surgeons’ experience and refined case selection, the trends in reduced conversion may vary by the surgical approach used. A previous study on laparoscopic CRC surgery reported a decrease in conversion rates between 2011 and 2015 from 11.8% to 8.6% in colon cancer and from 13% to 8% in rectal cancer. 13 We assumed there is a trend of reduced conversion from robotic-assisted to open proctectomy for rectal cancer over time; however, the pattern and significance of this trend may vary among different groups of patients. Therefore, the present study aimed to assess the time-based trends in conversion from RAP to laparotomy according to patient demographics, treatment facility, insurance types, and type of proctectomy.
Patients and Methods
Study design and data source
This study was a retrospective cohort analysis of the trends in conversion from RAP in patients with stage I–III rectal cancer to open surgery. The study also comprised a case-control analysis of the risk factors for conversion to laparotomy. Cases were defined as patients converted from robotic-assisted to open proctectomy, and controls were patients who had their RAP completed without conversion. Conversion was defined using the National Cancer Database (NCDB) parameter “Approach—Surgery of the Primary Site at this Facility” with code 1 indicating the completion of RAS without conversion and code 2 indicating converted robotic surgery.
Data were obtained from the NCDB from 2010 to 2021. The NCDB includes data from more than 1500 United States hospitals that are accredited by the Commission on Cancer (CoC) and is considered a joint project of the CoC of the ACS and the American Cancer Society. Given the retrospective nature of the study that used de-identified patients’ data from a national database, the study was not considered human subject research, and therefore ethics committee approval was not required. The de-identified data used in the study are derived from the NCDB and its participating hospitals that are not responsible for the statistical validity of the analysis or the conclusions of the study.
Selection criteria
Patients with stage I–III rectal adenocarcinoma who underwent RAP were included. The disease stage was defined according to the collective TNM stage parameter, and adenocarcinomas were identified using the International Classification of Diseases for Oncology, 3rd Edition (ICDO-3) codes: 8140/3, 8480-8481/3, and 8490/3. To minimize sampling and selection bias, we included consecutive patients from a large national database.
The following patients were excluded from the study: patients with stage IV disease or unknown stages; patients who did not undergo surgery, underwent local excision, or when the surgery type was not specified, patients who underwent open or laparoscopic proctectomy, or when the surgical approach was unknown.
Data collected
Data collected from the NCDB and used in the present study included patients’ characteristics (age, sex, race, Charlson comorbidity index score, insurance status, residence area, and facility type); tumor characteristics (TNM stage, size, and histological type); treatments (neoadjuvant therapy and type of proctectomy). Types of surgery included low anterior resection (LAR), abdominoperineal resection (APR), pull-through hand-sewn coloanal anastomosis with sphincter preservation, and pelvic exenteration (PE), which includes proctectomy with resection in continuity with other organs. The main study outcome was conversion from robotic-assisted to open proctectomy.
Data analysis
EZR (Easy R) (version 1.55) and R software (version 4.1.2) were used to perform the statistical analyses. Continuous data were expressed as the median and interquartile range (IQR) and were analyzed with the Mann–Whitney U. Categorical data were expressed as counts and absolute proportions and were analyzed using the Fisher exact or chi-square test. A complete case analysis was used to handle missing data. Factors significantly (P < .05) associated with conversion to laparotomy in the univariate association analysis were selected for a binary logistic regression multivariable analysis. The area under the curve (AUC) of the model was used to assess the discriminatory ability of the model. Multicollinearity between the predictors included in the model was evaluated using the variance inflation factor (VIF), where a VIF >10 indicated substantial multicollinearity. The changes in the rates of conversion to laparotomy over time were plotted as line graphs and the significance of the trend in change was estimated with the Cochran-Armitage trend test. P values <.05 were considered significant.
Results
Cohort description
After screening the records of 250,169 patients with rectal cancer, 23,644 patients were included (Fig. 1). The median age of patients was 60 (IQR: 51, 69) years, and 18.9% of patients were aged <50 years. Most patients were male (62.3%), White (85.7%), with a Charlson score of 0 (76.7%), lived in metropolitan areas (82%), and had private insurance (53.7%). More than one-third (38.1%) of patients were treated in academic/research programs. Overall, 18.6% of rectal cancers were of stage I, 29% of stage II, and 52.4% of stage III. Mucinous and signet-ring cell carcinomas accounted for 3.1% and 0.4% of cancers, respectively. The median tumor size was 40 (IQR: 27, 55) mm, and 38.3% of rectal cancers were ≥5 cm. Neoadjuvant radiation therapy was given to 72.8% of patients. The most performed surgery was LAR (69.8%), followed by APR (21.2%), pull-through coloanal anastomosis (7.2%), and PE (1.7%) (Table 1).
Included patients.
Characteristics of Patients Converted to Open Surgery Compared to Patients Who Were Not Converted
Bold text in P value column indicates statistical significance.
Conversion to open surgery
Conversion to open surgery was done in 1280 patients (5.4%, 95% CI: 5.1%–5.7%). Patients whose operations were converted to laparotomy more often were aged ≥50 years (84.9% versus 80.8%), were male (70% versus 61.8%), were Black (8.8% versus 7%), had a Charlson score >1 (9.1% versus 6.8%), had urban residence (18.6% versus 15.8%), and were treated at comprehensive community cancer programs (41.5% versus 35%). Lower conversion rates were observed in patients with private insurance (49.4% versus 53.9%) and patients with a median household income ≥ $63,000 (28.5% versus 35.9%). Higher rates of conversion were noted in rectal cancers ≥5 cm (44.5% versus 38%), and with neoadjuvant therapy (75.6% versus 72.7%), APR (22.7% versus 21.1%), and PE (2.6% versus 1.7%) (Table 1).
Multivariable regression analysis showed that the independent predictors for increased conversion were higher Charlson scores (score 1—OR: 1.32, P = .005; score 2—OR: 1.67, P = .002; score 3—OR: 1.74, P = .007), community cancer programs (OR: 1.99, P < .001), comprehensive community cancer programs (OR: 1.37, P < .001), household income of $48,000–$62,999 (OR: 1.29, P = .014) and of $38,000–$47,999 (OR: 1.45, P < .001), tumors ≥5 cm (OR: 1.35, P < .001), and PE (OR: 2.19, P = .001). Conversely, female sex (OR: 0.72, P < .001), pull-through coloanal anastomosis (OR: 0.69, P = .037), and the year of diagnosis of rectal cancer (OR: 0.92, P < .001) were independently associated with a lower likelihood of conversion (Table 2). The AUC of the model was 0.639 (95% CI: 0.618–0.66), indicating an acceptable discriminatory ability of the model. The VIF of all predictors ranged between 1.01 and 2.0, suggesting no significant collinearity.
Predictors for Conversion from Robotic-Assisted Proctectomy to Open Surgery
Bold text in the P value column indicates statistical significance.
OR, odds ratio; CI, confidence interval.
Time-based trends in conversion
There was a significant linear trend of decreased conversion to open surgery over time, with a conversion rate of 3.9% in 2021 compared to 10.4% in 2010 (reduction extent: 62.5%, P < .001). The year 2018 marks the cutoff for a significant reduction in conversion rates (sensitivity: 67.2% and specificity: 40.8%). The trend of reduced conversion was noted in different subgroups, yet with varying reduction extents and significance levels.
The reduction in conversion rates was significant in patients ≥50 years (10.4% to 3.7%, P < .001) but not in patients <50 years (10.4% to 5.2%, P = .838). Both males and females had a significant trend of reduced conversion; however, the reduction was more noted in males (12.3% to 4.1%) than in females (7.6% to 3.6%). While the trend of reduction in conversion was significant in White (10.3% to 4.1%, P < .001) and Asian patients (8.3% to 2.7%, P = .047), it was not significant in Black patients (16.7% to 3.5%, P = .358). The rate of conversion in 2010 was higher in Hispanic than in non-Hispanic patients (14.8% versus 10.4%); however, both had similar conversion rates in 2021 (3.8% versus 4.1%) (Fig. 2).
Conversion trends—demographics.
While a significant linear trend of reduced conversion was noted in patients with a Charlson score of ≤1; patients with a score >1 had a nonsignificant, nonlinear trend (P = .16). A significant trend of reduced conversion was noted across all facility types (academic/research programs [13% to 3.9%, P < .001], community cancer programs [25% to 4.7%, P = .022], comprehensive community cancer programs [7.9% to 4.3%, P < .001], and integrated network cancer programs [7.6% versus 3%, P < .001]) and all insurance types except with governmental insurance other than Medicaid and Medicare (Fig. 3).
Conversion trends—facility and insurance.
All disease stages showed a significant trend of reduced conversion, which was more noted in stage III (12.8% to 4.2%, P < .001) than in stage II (7.9% versus 3.8%, P < .001). In 2010, tumors ≥5 cm in size had higher conversion rates than tumors <5 cm (12.4% versus 9.8%); however, they showed a significant trend of reduced conversion with conversion rates in 2021 of 5.7% versus 3.3% for tumors <5 cm. The use of neoadjuvant therapy did not affect the significant trend in conversion (neoadjuvant: 10.3% to 4.2% versus no neoadjuvant: 9.8% to 3.3%). Conversely, the type of surgery had a remarkable impact, as a significant reduction in conversion was noted with LAR (10.1% to 3.6%, P < .001) and pull-through coloanal anastomosis (13.3% to 4.9%, P < .001) but not with APR (9.6% versus 5.1%) or PE, which did not have a linear trend of reduced conversion (Fig. 4).
Conversion trends—disease stage, tumor size, and treatment.
The largest reductions in conversion rates (>80% reduction) were noted in patients living in urban and rural areas, patients treated at community cancer programs, patients without insurance, and patients undergoing PE. Conversely, the smallest reductions (<60%) were observed in patients <50 years, females, patients with a Charlson score ≥1, patients living in metropolitan areas, patients treated at comprehensive community cancer programs, patients with Medicaid or private insurance, patients with stage II or T4 disease, patients with tumors ≥5 cm, and patients who received neoadjuvant therapy and underwent APR. Table 3 summarizes the percent of reduction in conversion rates from 2010 to 2021 with significance level.
Percent Reduction in Conversion Rates (2010–2021) in Different Subgroups
Bold text in P value column indicates statistical significance.
Discussion
The present study found that conversion from RAP to laparotomy has decreased over time following a statistically significant linear trend, except for young and Black patients and patients undergoing APR or PE. Male sex, significant comorbidities, low household income, large tumors, and PE were independently associated with increased odds of conversion to open surgery. Overall, conversion from RAP to open surgery linearly decreased from 10.4% in 2010 to 3.9% in 2021, amounting to a 62.5% reduction. This remarkable reduction in conversion rates might be attributed to increasing experience with RAS, improvement in the technique, and the introduction of more advanced and better platforms and instruments, including simulators and dual consoles that may help hasten the learning curve.14–16 Another potential explanation for the reduced conversion rates is the careful selection of patients for RAS, as a recent study 14 highlighted several disparities in the use of RAP for rectal cancer and found that multi-visceral resections such as PE were less likely to be performed by RAS, consistent with PE being a significant risk factor for conversion in the present study.
Although most patients showed a significant linear trend of reduced conversion, this linear trend was not observed in some patients. Patients younger than 50 showed a remarkable drop in conversion rates in the early stage of the study period; however, conversion rates increased in the last year. While there is no clear explanation for this finding, we may assume that the presentation of early-onset patients with advanced rectal cancer 17 may have factored into this inconsistent trend. Furthermore, Black patients and patients with government insurance other than Medicare and Medicaid did not have a significant trend in reduced conversion. This finding may be explained by the access challenges and treatment disparities concerning racial and social determinants. It has been reported that access to robotic surgery in Black patients is 15% less than in White patients, whereas lack of insurance decreased access by 39%. 16 It is worth noting that despite the overall reduction in conversion rates in Black patients, the trend curve fluctuated over time, implying inconsistent patterns of care. The significant trend of reduced conversion was noted in both sexes. Although the male sex is associated with more technically demanding rectal resections 18 and higher conversions, as shown by a higher conversion rate than females in 2010, conversion in males followed a similar linear trend to females, reaching a similar rate in 2021.
The treating facility and insurance status are important factors to be considered when assessing treatment outcomes. All facility and insurance types were associated with a significant linear trend of reduced conversion, suggesting comparable care patterns and surgeons’ experience. It is important to highlight the striking reduction in conversion rates for community cancer programs and uninsured patients in the first 3 years of the study period, which then followed the trend line for the entire cohort. This observation may imply better access to high-quality surgical care and experienced robotic surgeons at community cancer programs, even for patients who did not have insurance.
While the disease stage did not affect the linear trend of reduced conversion, tumor size was impactful. Rectal cancers larger than 5 cm initially had higher conversion rates than the rates decreased over time yet remained higher than the conversion rates for small tumors. Dissection and radical resection of large CRCs (>5 cm) might be more challenging and may be associated with a higher incidence of positive resection margins up to 10%; however, minimally invasive surgery remains a viable option for large tumors despite a potentially higher conversion rate. 19
Neoadjuvant therapy may be associated with a more difficult rectal resection because of the ensuing fibrosis. Nonetheless, patients who received neoadjuvant therapy in our study had a similar trend of reduced conversion to patients who did not receive neoadjuvant therapy. Conversely, the type of resection had a significant impact on the trend of reduced conversion, as it was significant in LAR but not in APR and PE. PE showed a non-linear trend of reduced conversion, which drastically dropped from 20% to 33.3% in the first 4 years of the study period to 6%–11% in the last 4 years. While this trend is optimistic, further study of this inconsistent trend in PE is needed to decipher its causes and help improve future outcomes.
The present study is the first dedicated analysis of trends in conversion from robotic-assisted to open proctectomy. The large number and multicentric nature of data used may help reduce selection and sampling bias and improve external validity. The stratification of conversion trends according to patient and tumor characteristics and treatment factors can help understand treatment gaps and disparities. Based on the study findings, we recommended improving access to high-quality RAS in accredited rectal cancer centers to high-risk patients, including early-onset and Black patients, patients with significant comorbidities, patients with large tumors, and patients undergoing APR or PE. Furthermore, preoperative optimization and prehabilitation of high-risk patients may help decrease the likelihood of conversion to open surgery.
Limitations of the current study include the retrospective nature of data, which is associated with selection bias. An important limitation is the lack of standard definition and criteria for conversion to open surgery, which may have varied across the different institutions contributing data to the NCDB. 20 Only CoC-accredited centers contribute data to the NCDB; therefore, the trends noted in the study may not extent to other nonaccredited hospitals. The study did not cover the years before 2010, as the NCDB started to capture data on the surgical approach for proctectomy in 2010. Another important limitation is the lack of information on the case volume of the hospitals and surgeons as high-volume centers could have been associated with more notable trends of reduced conversion.
Conclusions
Unplanned conversion from robotic-assisted to open proctectomy showed a linear trend of reduction over time. This trend was statistically significant in all patient and treatment groups except in patients <50 years old, Black patients, patients with significant comorbidities, and patients undergoing APR or PE. Male sex, significant comorbidities, low household income, large tumors, and PE were independently associated with increased conversion.
Footnotes
Ethics Approval
Given the retrospective nature of the study that used de-identified patients’ data from a national database, the study was not considered human subject research, and therefore ethics committee approval was not required.
Data Availability
Upon reasonable request from first author.
Disclosure Statement
S.D.W. is a consultant for Activ Surgical, Arthex, Baxter, Becton, Dickinson and Co., Intuitive Surgical, OstomyCure, Takeda, and Virtual Ports, has consulting agreements with stock options for consulting with GI View, OstomyCure, and Virtual Ports, is the Data Safety Monitoring Board chair of Polypid, and receives royalties from Intuitive Surgical, Karl Storz Endoscopy America Inc., and Unique Surgical Solutions, LLC. Dr. Emile is a consultant for Becton, Dickinson and Co. Drs. N.H., Z.G., and E.S. do not report any conflicts of interest.
Funding Information
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.