Journal Logo

ORIGINAL ARTICLES

The Early (2009–2017) Experience With Robot-assisted Cholecystectomy in New York State

Hoffman, Aaron B. MD, FACS; Myneni, Ajay A. MBBS, PhD, MPH; Towle-Miller, Lorin M. MA; Karim, Syed A. MD; Train, Arianne T. DO, MPH; Burstein, Matthew MD; Schwaitzberg, Steven D. MD, FACS; Noyes, Katia PhD, MPH∗,‡

Author Information
doi: 10.1097/SLA.0000000000004932

Abstract

Cholecystectomies are among the most commonly performed general surgical procedures with approximately one million performed annually in the United States.1 Soon after its introduction in the late 1980s, laparoscopic cholecystectomy became the standard of care for the majority of gallbladder procedures and is associated with reduced length of stay and hospital costs.1–3 Potential complications of cholecystectomy include hemorrhage, infection, and injury to surrounding structures.3 Although rare, iatrogenic bile duct injury (BDI) is the most serious preventable complication with a reported incidence of 1 to 2 per 1000 patients undergoing cholecystectomy.4,5 BDI is associated with disastrous consequences in terms of morbidity, mortality, and high health care costs.4,6,7 Management of BDI often involves the need for additional procedures ranging from minor interventions to major reconstructive procedures, often performed by experienced hepatobiliary surgeons.8,9

Over the past decade, use of robotic assistance for laparoscopic cholecystectomies has increased dramatically.10,11 Although robotic assistance may offer some technological advantages,12,13 robot-assisted cholecystectomies (RACs) are associated with higher procedural costs, specialized operating room staffing, and longer operating times compared to traditional laparoscopic cholecystectomies (LCs).14–21

Previous studies comparing RAC versus LC approach often used data from single early-adopter RAC institutions performed by expert Robotic surgeon-researchers. A frequently cited study specifically excluded novice robotic surgeons who had performed <125 RACs.22 This methodology artificially eliminates the learning curve, and led those studies to conclude that RAC had a superior safety profile and fewer complications compared to LC.12,14,16,18 In contrast, the studies examining larger inpatient datasets from large hospital systems and state and national discharge databases showed higher complication rates and costs associated with RAC compared to LC.17,20,21

To better understand real-world utilization of RAC and identify factors that may be affecting patient outcomes post-cholecystectomy, we evaluated recent trends in the use of minimally invasive surgical (MIS) cholecystectomy in New York State (NYS) using 2009–2017 all-age all-payer state hospital discharge database. In addition, we assessed long-term health and economic outcomes following the index surgery and changes in provider volumes.

METHODS

Study Sample

We analyzed data of patients undergoing inpatient and ambulatory surgery from the NYS's State Planning and Research Cooperative System (SPARCS) starting from 2009 when the explicit coding for robotic assisted laparoscopic surgery was first introduced. Using ICD and CPT codes23,24 (see Table, Supplemental Table 1, http://links.lww.com/SLA/D135), we identified 325,492 patients who underwent MIS cholecystectomy (Fig. 1). Exclusions included patients <18 years of age (1.6%); non-NYS residents due to potentially missing follow-up data (1.9%); any duplicate patient records (0.6%), those with liver, biliary tract, or pancreatic malignancies diagnosed up to 1 year after the date of cholecystectomy (0.7%); patients with preoperative diagnoses of cirrhosis of the liver or congenital choledochal cysts (1.0%); and those who underwent incidental cholecystectomies (cholecystectomies performed in addition to or as a part of another surgical procedure) (2.9%) and length of stay >1 year (n = 1). Our final analytic sample consisted of 299,306 patients (Fig. 1). The study was approved by the University at Buffalo's Institutional Review Board, and a Data Use Agreement with NYS Department of Health guided the use of the SPARCS data.

FIGURE 1
FIGURE 1:
Study sample flow chart showing initial sample, exclusions and final analytic sample. Up to 1 year from date of index surgery, if present on admission, if cholecystectomy was performed in addition to or as a concomitant procedure along with other gastrointestinal surgery.

Patient Characteristics and Study Outcomes

We examined preoperative patient characteristics including age, sex, race, county of residence (urban or rural as defined by US Census Bureau),25 distance travelled to receive surgery, and insurance type. We also examined comorbidities at the time of surgery including morbid obesity, diabetes mellitus, essential hypertension, acute cholecystitis, chronic cholecystitis, and symptomatic cholelithiasis. We also used Elixhauser comorbidity index, a numeric score to assess patients’ comorbidity burden based on 30 categories defined by ICD diagnosis codes.26

Principal outcomes included BDIs, conversion to open procedure, intraoperative cholangiography, and perioperative or postoperative interventions on the bile duct reported up to 1 year from the date of index LC/RAC procedure. The ICD-9, ICD-10, and CPT codes used to define the outcome variables are presented in Supplemental Table 1, http://links.lww.com/SLA/D135. Bile duct interventions were classified into 3 categories—endoscopic, minor, and major. Minor bile duct interventions included repair procedures other than anastomoses to the intestine. Major interventions included procedures involving anastomoses of biliary ducts to the intestine.

Secondary outcomes included median hospital length of stay; postoperative inpatient readmissions and unplanned visits to emergency department or ambulatory hospital centers up to 30 days from LC/RAC; and 30-day (except as noted) postoperative outcomes including acute pancreatitis (≤14 days), postoperative hemorrhage, retained stones, percutaneous abdominal drainage, sepsis, postoperative infection (including surgical site infections), myocardial infarction, pneumonia, urinary tract infection, pulmonary embolism, deep vein thrombosis, requirement for venous catheterization, blood transfusion, parenteral antibiotics, and other post procedural digestive system complications. We examined the 30-day postoperative outcomes as individual complications as well as a composite summary if patients had at least one of the complications. We additionally examined incisional hernia and death up to 1 year from the date of LC/RAC. The table Supplemental Table 1, http://links.lww.com/SLA/D135 lists the ICD and CPT codes used for the outcome variables.

Hospital charges were used as a proxy to estimate healthcare burden after surgery. We evaluated the hospital charges for the LC/RAC admission, as well as each patient's cumulative hospital charges up to 1 year from their LC/RAC procedure.

Temporal Trends of Complications

We examined temporal trends of the proportion of patients with any secondary complications following LC/RAC. Due to the lower sample size of RAC patients, any secondary complications (2009–2011) were too low to derive meaningful conclusions and were not presented.

Estimating Surgeon Experience

We tracked all surgeons in NYS who performed at least 1 RAC between 2009 and 2017. Using 2009 as a screening period for collecting existing RAC surgeons, “new RAC surgeons” were extracted from 2010 to 2017 when their NYS license number or national provider identifier first appeared as the operating surgeon for an RAC procedure. We subsequently plotted the number of new and total RAC surgeons for each year between 2009 and 2017.

Statistical Analyses

To examine differences between patients who underwent RAC versus LC, Fisher exact tests and Wilcoxon rank sum tests were conducted on the preoperative patient characteristics, principal outcomes, and secondary outcomes. To make the RAC and LC samples more comparable, we performed propensity score matching (PSM) based on age, sex, race, insurance, preoperative comorbidities, and Elixhauser Index. We then reconducted all analyses on the matched subsample.

We also utilized multiple logistic regression models to estimate effect of procedure type (RAC vs LC) on postoperative complications, including any bile duct complication, conversion to open procedure, postoperative encounters including inpatient admissions and return visits to emergency department or ambulatory surgery centers, and 30-day postoperative complications using the PSM samples. The multivariate models were adjusted for age, sex, race, morbid obesity (yes/no), distance travelled for LC/RAC, type of insurance, patient rural status, preoperative diagnosis of acute cholecystitis, and Elixhauser Index. Weighted least squares regression was used to test for temporal trends in secondary complication rates, where weights were based on the number of cholecystectomies performed. A plot was generated to display the trends and regression results. We calculated the slope (x) and associated P values for the fitted regression lines in the plots and we compared the slopes between the LC/RAC groups. All analyses were conducted using SAS 9.4 (Cary, NC).

RESULTS

Patient Characteristics

Our analytic sample (Fig. 1) consisted of 299,306 patients who underwent MIS cholecystectomy in NYS between 2009 and 2017. Of those, 298,188 (99.6%) patients underwent LC and 1118 (0.4%) patients underwent RAC (Table 1). Compared to those who underwent LC, RAC patients were older (mean age of 51.4 years vs 49.4 years among LC patients, P < 0.01), had a higher proportion of African Americans (14.7% vs 10.2%, P < 0.01), and were more likely covered by Medicaid or Medicare (49.5% vs 41% of LC patients). Both rural and urban patients travelled significantly longer distances to receive RAC compared to LC surgery. At the time of the index surgery, a greater proportion of RAC compared to LC patients were diagnosed with morbid obesity (12.2% vs 5.7%; P < 0.01), diabetes (18.1% vs 12.3%; P < 0.01), and essential hypertension (35.3% vs 29.3%; P < 0.01). RAC patients were less likely than LC patients to present with chronic cholecystitis or symptomatic cholelithiasis (72.5% vs 81.0%, P < 0.01). The table Supplemental Table 2, http://links.lww.com/SLA/D135 shows that the subsampled RAC and LC patients were adequately matched on preoperative characteristics.

TABLE 1 - Preoperative Characteristics of Patients Who Underwent LC and RAC in New York State (2009–2017), N = (299,306)
Variable LC, N (%) RAC, N (%) P
Total 298,188 (99.6) 1118 (0.4)
Age, y, mean (SD) 49.4 (17.5) 51.4 (17.7) <0.01
Sex, females 211,486 (70.9) 791 (70.8) 0.92
Race <0.01
 White 191,217 (65.3) 665 (63.5)
 Black/African American 29,911 (10.2) 154 (14.7)
 Other 71,661 (24.5) 229 (21.8)
Patient location (missing n = 7546)
 Urban county 261,182 (89.9) 999 (90.6) 0.45
 Rural county 29,477 (10.1) 104 (9.4)
Median distance travelled distance (miles)
 Urban county <1.0 6.0 <0.01
 Rural county 27.4 53.9 <0.01
Patient insurance type <0.01
 Medicaid 69,400 (23.3) 312 (28.0)
 Medicare 52,689 (17.7) 240 (21.5)
 Private/Other 160,424 (53.9) 535 (47.9)
 Uninsured 14,979 (5.0) 29 (2.6)
Preoperative comorbidities
 Morbid obesity 16,993 (5.7) 137 (12.2) <0.01
 Diabetes 36,543 (12.3) 203 (18.1) <0.01
 Essential hypertension 87,302 (29.3) 395 (35.3) <0.01
 Elixhauser readmission score, mean (SD) 3.2 (7.6) 4.7 (8.8) <0.01
 Acute cholecystitis 37,817 (12.7) 161 (14.4) 0.09
 Chronic cholecystitis or symptomatic cholelithiasis 241,556 (81.0) 810 (72.5) <0.01
LC indicates laparoscopic cholecystectomy; RAC, robot-assisted cholecystectomy.
P value was derived from Fischer exact tests or Wilcoxon rank sum tests.
Urban counties are those with >50% population living in urban areas and rural counties are those with >50% population living in rural areas, based on 2010 Census data.25
Presenting with acute or acute and chronic cholecystitis at the time of surgery.

Patient Outcomes

RAC patients had significantly higher rates of conversion to open procedure compared to LC patients (4.9% vs 2.8%; P < 0.01). Intraoperative cholangiography was more often performed among LC patients (8.7% vs 6.2% among RAC patients, P < 0.01). A higher proportion of RAC patients were diagnosed with BDI within 1 year from the index procedure compared to LC patients (1.3% vs 0.4%; P < 0.01) (Table 2). Other bile duct-related complications including bile duct fistula, bile duct obstruction, as well as postoperative cholangitis postoperative seroma/biloma did not differ among RAC versus LC patients.

TABLE 2 - Intraoperative and Postoperative Outcomes in Patients Who Underwent LC and RAC in New York State (2009–2017), N = 299, 306
Study Outcomes LC, N (%) RAC, N (%) P
No. of procedures 298,188 (99.6) 1118 (0.4)
Primary outcomes
 Conversion to open procedure 8369 (2.8) 55 (4.9) <0.01
 Intraoperative cholangiogram 25,888 (8.7) 69 (6.2) <0.01
Bile duct complications
  Bile duct injury 1316 (0.4) 14 (1.3) <0.01
  Bile duct fistula 84 (0.03) 1 (0.09) 0.27
  Bile duct obstruction 2127 (0.7) 13 (1.2) 0.10
  Postoperative cholangitis 3928 (1.3) 15 (1.3) 0.90
 Bile duct interventions
  Endoscopic interventions 8758 (2.9) 34 (3.0) 0.79
  Nonendoscopic interventions 1187 (0.4) 14 (1.3) <0.01
   Minor interventions 927 (0.3) 7 (0.6) 0.09
   Major interventions 260 (0.1) 7 (0.6) <0.01
Secondary outcomes
 Length of stay, days, median (IQR)
  Among all patients 1 (0–3) 3 (1, 5) <0.01
  Excluding patients with complications, median (IQR) 0 (0–2) 2 (1, 4) <0.01
 Postoperative patient encounters (≤30 days from index surgery) 67,676 (22.7) 315 (28.2) <0.01
 Inpatient readmissions 13,164 (4.4) 82 (7.3) <0.01
 Ambulatory surgery/emergency department visits 58,684 (19.7) 265 (23.7) <0.01
 30-day Postoperative complications 45,669 (15.3) 285 (25.5) <0.01
  Acute pancreatitis (≤14 days from date of LC/RAC) 23,365 (7.8) 146 (13.1) <0.01
  Postoperative haemorrhage 2538 (0.9) 18 (1.6) 0.01
  Postoperative retained stones 664 (0.2) 1 (0.1) 0.53
  Percutaneous abdominal drainage 1721 (0.6) 12 (1.1) 0.04
  Sepsis 4829 (1.6) 21 (1.9) 0.48
  Postoperative infection 1812 (0.6) 6 (0.5) 1.00
  Myocardial infarction 1555 (0.5) 8 (0.7) 0.40
  Pneumonia 3648 (1.2) 19 (1.7) 0.17
  Urinary tract infection 5911 (2.0) 27 (2.4) 0.28
  Pulmonary embolism 1575 (0.5) 6 (0.5) 0.84
  Deep vein thrombosis 735 (0.3) 2 (0.2) 1.00
  Venous catheterization 1555 (0.5) 4 (0.4) 0.67
  Blood transfusion 4750 (1.6) 31 (2.8) <0.01
  Parenteral antibiotics 3876 (1.3) 38 (3.4) <0.01
Other GI complications ± 2699 (0.9) 28 (2.5) <0.01
incisional hernia§ 2829 (1.0) 26 (2.3) <0.01
Death§ 1811 (0.6) 9 (0.8) 0.34
GI indicates gastrointestinal gastrointestinal; IQR, inter-quartile range; LC, laparoscopic cholecystectomy; RAC, robot-assisted cholecystectomy.
P value was derived from Fischer exact tests or Wilcoxon rank sum tests.
Up to 1 year from date of index procedure (LC/RAC).
Sample that excluded patients with any primary or secondary complications ± includes ICD codes for postprocedural hepatic failure, postprocedural hepatorenal syndrome, and other post procedural GI complications.
§Up to 1 year from date of index procedure (LC/RAC).

Among interventions to treat bile duct complications, endoscopic interventions did not differ among RAC and LC patients (2.9% vs 3.4% P = 0.79). Minor surgical bile duct repair interventions also did not differ between RAC and LC patients (0.6% vs 0.3%, P = 0.09) However, major reconstructive surgeries were much higher among RAC versus LC patients (0.6% vs 0.1%, P < 0.01). The results for the primary outcomes remained similar in the propensity score matched sample (Supplemental Table 3, http://links.lww.com/SLA/D135).

RAC patients on average spent a longer time (median of 3 days) in the hospital following surgery compared to LC patients (median of 1 day). Inpatient readmissions and unplanned visits to emergency department or ambulatory surgery centers within 30 days of cholecystectomy were more likely among RAC compared to LC patients (P < 0.01). A greater proportion of RAC patients encountered acute pancreatitis (P<0.01), postoperative hemorrhage (P = 0.01), required percutaneous abdominal drainage (P = 0.04), blood transfusions (P < 0.01), and parenteral antibiotics (P < 0.01) compared to LC patients. Incisional hernia was also more common among RAC compared to LC patients (P < 0.01). Postoperative sepsis, myocardial infarction, pneumonia, and urinary tract infection were higher among RAC versus LC patients, but the differences were not statistically significant. Mortality up to 1 year did not differ between RAC and LC patients. In the PSM subsample, the differences in postoperative hemorrhage, blood transfusion, and percutaneous abdominal drainage were no longer statistically significant. However, acute pancreatitis, use of parenteral antibiotics, and incisional hernia remained higher among RAC versus LC patients (see table, Supplemental Table 3, http://links.lww.com/SLA/D135).

Multivariable Analyses

After controlling for patient demographics and comorbidities on the PSM samples, RAC patients had 175% greater odds of having postoperative bile duct interventions [adjusted odds ratio (aOR) 2.75, 95% confidence intervals (CI): 1.55–4.88]. Cases that started as RAC had 69% greater odds of conversion to an open procedure (aOR 1.69, 95% CI: 1.26–2.28) compared to cases that started as LC. RAC patients had 49% greater odds of having at least one 30-day postsurgical complication (aOR 1.49, 95% CI: 1.28, 1.73). Additionally, RAC patients had significantly higher odds of an inpatient readmission as compared to LC patients (aOR 1.40, 95% CI: 1.09–1.79), but the odds of an unplanned visits to emergency department or ambulatory surgery centers was not significantly different (Table 3).

TABLE 3 - Association of Select Outcome Variables Among Patients Who Underwent LC and RAC in New York State (2009–2017) After Propensity Score Matching, N = (32,426)
Outcome LC, N (%) RAC, N (%) RAC vs LC aOR (95% CI)
Bile duct interventions 143 (0.46) 13 (1.2) 2.75 (1.55–4.88)
Conversion to open procedure 1051 (3.4) 51 (4.9) 1.69 (1.26–2.28)
Postoperative encounters 7434 (23.7) 297 (28.4) 1.12 (0.97–1.29)
Inpatient readmissions 1506 (4.8) 75 (7.2) 1.40 (1.09–1.79)
Ambulatory surgery/emergency department visits 6426 (20.5) 252 (24.1) 1.09 (0.94–1.26)
30-day Postoperative complications§ 4596 (14.6) 203 (19.4) 1.49 (1.28–1.74)
LC indicates laparoscopic cholecystectomy; RAC, robot-assisted cholecystectomy.
Adjusted for distanced travelled to undergo LC/RAC.
Includes endoscopic and non-endoscopic bile duct interventions encountered ≤1 year from LC/RAC.
Up to 30 days from LC/RAC.
§Includes acute pancreatitis (≤14 days from date of LC/RAC), postoperative hemorrhage, retained stones, abdominal drainage, sepsis, infections, myocardial infarction, pneumonia, deep vein thrombosis, urinary tract infection, pulmonary embolism, requirement for venous catheterization, blood transfusion, parenteral antibiotics, or other post procedural complications.

Hospital Charges

Compared to LC patients, RAC patients faced significantly higher hospital charges for the index admission (difference in mean = $16,561, P < 0.01) as well as higher cumulative 12-month charges from date of index procedure (difference in mean = $25,373, P < 0.01)

Trends of Complications and Physician Experience

Figure 2 shows the proportion of LC and RAC patients with any secondary complications by year of index procedure between 2012 and 2017. Among LC patients, secondary complications showed a decreasing trend (x = −0.01, P = 0.03), whereas there was no discerning trend among RAC patients (x = 0.003, p = 0.96) and there was no difference in trends between LC and RAC patients (P = 0.82)

FIGURE 2
FIGURE 2:
Proportion of any secondary complications by year among LC and RAC patients. Any secondary complications included acute pancreatitis (≤14 days postoperative), 30-day postoperative outcomes including hemorrhage, retained stones, percutaneous abdominal drainage, other GI complications, sepsis, postoperative infection, myocardial infection, pneumonia, urinary tract infection, pulmonary embolism, deep vein thrombosis, venous catheterization, blood transfusion, parenteral antibiotics, other GI complications (includes postprocedural hepatic failure, postprocedural hepatorenal syndrome, and other post procedural digestive system complications), as well as incisional hernia and death which were followed up to 1 year from date of index LC/RAC procedure. The plot shows fitted regression lines weighted by sample size. The inset in the plots shows the calculated slopes (x) for the LC and RAC groups as well as the comparison of the slopes and the associated P values. The table below the plot shows the frequencies of complications among LC/RAC patients. GI indicates gastrointestinal.

Surgeon Experience

We found that the number of surgeon performing RAC in NYS increased from 2 in 2009 to 70 in 2017 (Fig. 3A). Additionally, the proportion of surgeons performing RAC for the first time among total RAC surgeons decreased from 100% in 2009 to 22.8% in 2017 (Fig. 3B).

FIGURE 3
FIGURE 3:
MIS surgeons practicing in New York State between 2010 and 2017 in our analytic sample. A, The lighter gray bars represent total practicing MIS (LC and RAC) surgeons. The dark gray portion represents the proportion of MIS surgeons who performed RACs that year. B, the shaded gray bars represent total RAC surgeons who performed at least 1 procedure that year. The patterned portion of the bars represents the proportion of new RAC surgeons that appeared in our dataset for the first time.

DISCUSSION

Using 2009–2017 NYS all-payer statewide inpatient admission and ambulatory surgery database, we found that that the use of RAC in NYS remained below 1% of all minimally invasive cholecystectomies) and was limited primarily to large urban hospitals. Although the total number of NYS surgeons performing RAC steadily increased (2 in 2009 to 70 in 2017), the percentage new RAC surgeons concurrently decreased (100% in 2009 to 22.8% in 2017). We found a significantly higher rate of conversion to open procedures and postoperative complications following RAC compared to LC. Specifically, RAC patients had a higher rate of perioperative and postoperative bile duct injuries compared to LC, increased rates of postoperative interventions to manage those injuries, longer length of stay, and higher overall costs.

These patterns are similar to the experience of laparoscopic cholecystectomy when it was first introduced in the late 1980s and early 1990s, which resulted in a marked increase in bile duct injuries compared to open cholecystectomy.18,27–30 This was attributed to lack of laparoscopic surgical experience.27 As with any new technique, complication rates greatly decreased over time.31 Our results demonstrate that >30 years later, this decline still continues.

It is not clear whether the observed differences in complications between RAC and LC stem from the experience of individual surgeons or from institutional factors like availability of highly specialized staff, institutional training and coaching, academic culture, or referral patterns. Despite both RAC and LC having camera-based visualization, the 3-D visualization of RAC did not appear to mitigate the effect of RA inexperience on BDI rates. Murphy et al (2010) reported that patient selection rather than hospital and surgeon operating volumes may be more predictive of postoperative complications forllowing LCs.32

We also found that the proportion of RAC cases that were converted to open cholecystectomies in our study was almost twice as high compared to LCs (4.9% vs 2.8%, P < 0.01). Although we did not have information regarding the reasons for conversion, the most common reported reasons for conversion to open surgery are intraoperative hemorrhage, decreased visualization of the biliary tree, and intraoperative recognition of injury to the biliary tract.33 Our analysis also showed that intraoperative cholangiographies were performed less commonly during RACs versus LCs (6.2% vs 8.6%, P < 0.01) potentially resulting in higher BDI rates. We surmise there is a significant disincentive for the surgeon to leave the robotic console and scrub back into the case for both routine cholangiography and for anatomic identification of biliary structures in difficult cases. Routine cholangiography was previously shown to have protective effects against BDI.34 Another recent publication from a high volume RAC center also reported that use of indocyanine green reduced open conversions and improved postoperative clinical outcomes among patients.22

Resource utilization was also found to be greater in RAC versus LC. We found a higher median LOS among RAC patients compared to LC patients, which was persistent even after excluding patients that encountered any of the primary and secondary complications (conversion to open procedure, bile duct injuries or interventions, or any secondary complications) as well as in the PSM subsample. We hypothesize that as with any novel procedure, physicians tend to be more cautious postoperatively and might want to observe the patient for a longer time to avoid unexpected complications.

Our data demonstrate that BDI and other postsurgical complication rates have not yet converged between RAC and LC in the 8 years between 2009 and 2017 in a population of surgeons generally familiar with MIS approaches to gallbladder disease. These findings remind us there is no substitute for surgical experience and appropriate training and that any major change in technique can lead to a transient increase in complications.35 Hence, use of the robotic approach should not be considered a short cut to compensate for the lack of surgical expertise.

We anticipate that RAC complication rates will continue to decrease and eventually approach that of LC, as RAC training and competencies are fully incorporated into General Surgery residency curriculum. Given our findings, it may be prudent to avoid RAC as a proctored or an early case for a junior surgeon or a provider who is new to RA technology. Although having more variable anatomy, hernias have a better safety profile for making it through the early robotic learning curve.

Our study has several limitations including those inherent to use of an administrative database.36–38 It is possible that certain LCs performed with robotic assistance before 2009 were miscoded before supplemental ICD-9, ICD-10, and CPT codes to indicate the use of robotic assistance was universally adopted in NYS. However, this would likely have resulted in non-differential misclassification making our results more conservative. Another instance of miscoding based on ICD diagnosis codes could have been encountered in defining acute cholecystitis before the index LC/RAC procedure. Identifying cases with acute cholecystitis from administrative data represents a challenge because of lack of ICD coding specificity and variation in hospital coding practices.37 However, it is unlikely that this would have introduced any statistical bias because we used the same codes in our definitions for both LC and RAC groups. Our estimates of new RAC surgeons are based on an administrative database for NYS between 2009 and 2017. These estimates may not reflect the actual numbers because it does not account for a surgeon's experience during residency or fellowship training or during practice in other states. Although our analyses are based on a large population-based dataset, there were too few RAC cases in the earlier years for a statistical analysis of annual trends. Hence, our results should be reexamined using national patient datasets or later years when new data become available. A final complicating factor might be the interaction between the introduction of a new and complex technology and experience of the surgeon. It was previously demonstrated in a large dataset that younger surgeons had a threefold higher common bile duct injury rate compared to their more experienced counterparts.29 Despite the above limitations, we believe this population study makes an important contribution to the literature by describing state-wide variation in surgical practices and patient outcomes outside of academic medical centers.

In conclusion, our analyses have demonstrated that the early adoption period of RAC in NYS was associated with higher rates of complications, most importantly bile duct related injuries. To minimize these preventable complications, robotic training paradigms may need to be adjusted for novice robotic surgeons, even those with extensive laparoscopic experience. Full utilization of available safety technology when performing RAC would improve outcomes, even in less-experienced hands. Given the significantly higher BDI rate in this early series of RAC and the devastating effects of BDI, we recommend robotic hernia repair as the initial procedure of choice to gain familiarity with the robotic platform. Novice robotic surgeons should limit their use of RAC early in their learning curve to avoid mirroring of the early period of LC adoption for any longer than necessary.

“Those who cannot remember the past are condemned to repeat it.”—George Santayana.

REFERENCES

1. Steiner CA, Karaca Z, Moore BJ, et al. Surgeries in Hospital-Based Ambulatory Surgery and Hospital Inpatient Settings, 2014: Statistical Brief #223. Rockville (MD): Healthcare Cost and Utilization Project (HCUP) Statistical Briefs; 2006.
2. Altieri MS, Brunt LM, Yang J, et al. Early cholecystectomy (< 72 h) is associated with lower rate of complications and bile duct injury: a study of 109,862 cholecystectomies in the state of New York. Surg Endosc 2020; 34:3051–3056.
3. Hassler KR, Jones MW. Laparoscopic Cholecystectomy. Treasure Island (FL): StatPearls; 2019.
4. Barrett M, Asbun HJ, Chien HL, et al. Bile duct injury and morbidity following cholecystectomy: a need for improvement. Surg Endosc 2018; 32:1683–1688.
5. Slater K, Strong RW, Wall DR, et al. Iatrogenic bile duct injury: the scourge of laparoscopic cholecystectomy. ANZ J Surg 2002; 72:83–88.
6. Strasberg SM. Biliary injury in laparoscopic surgery: part 2. Changing the culture of cholecystectomy. J Am Coll Surg 2005; 201:604–611.
7. Strasberg SM. Biliary injury in laparoscopic surgery: part 1. Processes used in determination of standard of care in misidentification injuries. J Am Coll Surg 2005; 201:598–603.
8. Pesce A, Palmucci S, La Greca G, et al. Iatrogenic bile duct injury: impact and management challenges. Clin Exp Gastroenterol 2019; 12:121–128.
9. Schreuder AM, Busch OR, Besselink MG, et al. Long-term impact of iatrogenic bile duct injury. Dig Surg 2019. 1–12.
10. Sheetz KH, Claflin J, Dimick JB. Trends in the adoption of robotic surgery for common surgical procedures. JAMA Netw Open 2020; 3:e1918911.
11. Zaman JA, Singh TP. The emerging role for robotics in cholecystectomy: the dawn of a new era? Hepatobiliary Surg Nutr 2018; 7:21–28.
12. Ayloo S, Roh Y, Choudhury N. Laparoscopic versus robot-assisted cholecystectomy: a retrospective cohort study. Int J Surg 2014; 12:1077–1081.
13. Lanfranco AR, Castellanos AE, Desai JP, et al. Robotic surgery: a current perspective. Ann Surg 2004; 239:14–21.
14. Breitenstein S, Nocito A, Puhan M, et al. Robotic-assisted versus laparoscopic cholecystectomy: outcome and cost analyses of a case-matched control study. Ann Surg 2008; 247:987–993.
15. Higgins RM, Frelich MJ, Bosler ME, et al. Cost analysis of robotic versus laparoscopic general surgery procedures. Surg Endosc 2017; 31:185–192.
16. Huang Y, Chua TC, Maddern GJ, et al. Robotic cholecystectomy versus conventional laparoscopic cholecystectomy: A meta-analysis. Surgery 2017; 161:628–636.
17. Kaminski JP, Bueltmann KW, Rudnicki M. Robotic versus laparoscopic cholecystectomy inpatient analysis: does the end justify the means? J Gastrointest Surg 2014; 18:2116–2122.
18. Kane WJ, Charles EJ, Mehaffey JH, et al. Robotic compared with laparoscopic cholecystectomy: a propensity matched analysis. Surgery 2020; 167:432–435.
19. Khorgami Z, Li WT, Jackson TN, et al. The cost of robotics: an analysis of the added costs of robotic-assisted versus laparoscopic surgery using the National Inpatient Sample. Surg Endosc 2019; 33:2217–2221.
20. Pokala B, Flores L, Armijo PR, et al. Robot-assisted cholecystectomy is a safe but costly approach: a national database review. Am J Surg 2019; 218:1213–1218.
21. Sun N, Zhang J, Zhang C, et al. Single-site robotic cholecystectomy versus multi-port laparoscopic cholecystectomy: a systematic review and meta-analysis. Am J Surg 2018; 216:1205–1211.
22. Gangemi A, Danilkowicz R, Elli FE, et al. Could ICG-aided robotic cholecystectomy reduce the rate of open conversion reported with laparoscopic approach? A head to head comparison of the largest single institution studies. J Robot Surg 2017; 11:77–82.
23. Aguayo E, Dobaria V, Nakhla M, et al. National trends and outcomes of inpatient robotic-assisted versus laparoscopic cholecystectomy. Surgery 2020; 168:625–630.
24. Halbert C, Pagkratis S, Yang J, et al. Beyond the learning curve: incidence of bile duct injuries following laparoscopic cholecystectomy normalize to open in the modern era. Surg Endosc 2016; 30:2239–2243.
25. Ratcliffe M, Burd C, Holder K, et al. Defining Rural at the U.S. Census Bureau,” ACSGEO-1, U.S. Washington, DC: Census Bureau; 2016.
26. Elixhauser A, Steiner C, Harris D R, et al. Comorbidity measures for use with administrative data. Med Care 1998; 36:8–27.
27. A prospective analysis of 1518 laparoscopic cholecystectomies. The Southern Surgeons Club. N Engl J Med 1991; 324:1073–1078.
28. Mangieri CW, Hendren BP, Strode MA, et al. Bile duct injuries (BDI) in the advanced laparoscopic cholecystectomy era. Surg Endosc 2019; 33:724–730.
29. Schwaitzberg SD, Scott DJ, Jones DB, et al. Threefold increased bile duct injury rate is associated with less surgeon experience in an insurance claims database: more rigorous training in biliary surgery may be needed. Surg Endosc 2014; 28:3068–3073.
30. Vidovszky TJ, Smith W, Ghosh J, et al. Robotic cholecystectomy: learning curve, advantages, and limitations. J Surg Res 2006; 136:172–178.
31. van de Graaf FW, Zaimi I, Stassen LPS, et al. Safe laparoscopic cholecystectomy: a systematic review of bile duct injury prevention. Int J Surg 2018; 60:164–172.
32. Murphy MM, Ng SC, Simons JP, et al. Predictors of major complications after laparoscopic cholecystectomy: surgeon, hospital, or patient? J Am Coll Surg 2010; 211:73–80.
33. Sakpal SV, Bindra SS, Chamberlain RS. Laparoscopic cholecystectomy conversion rates two decades later. JSLS 2010; 14:476–483.
34. Altieri MS, Yang J, Obeid N, et al. Increasing bile duct injury and decreasing utilization of intraoperative cholangiogram and common bile duct exploration over 14 years: an analysis of outcomes in New York State. Surg Endosc 2018; 32:667–674.
35. Jayaraman S, Davies W, Schlachta CM. Getting started with robotics in general surgery with cholecystectomy: the Canadian experience. Can J Surg 2009; 52:374–378.
36. Guller U. Surgical outcomes research based on administrative data: inferior or complementary to prospective randomized clinical trials? World J Surg 2006; 30:255–266.
37. Murphy M, Alavi K, Maykel J. Working with existing databases. Clin Colon Rectal Surg 2013; 26:5–11.
38. Rabinstein AA. Administrative medical databases for clinical research: the good, the bad, and the ugly. Neurocrit Care 2018; 29:323–325.
Keywords:

bile duct injuries; laparoscopic cholecystectomy; population-based study; quality improvement; robot-assisted cholecystectomy; surgical education

Supplemental Digital Content

Copyright © 2021 Wolters Kluwer Health, Inc. All rights reserved.