Comparative analysis of opioid use between robotic and open pancreatoduodenectomy

Abstract

Background/Purpose:

Risk-stratified pancreatectomy clinical pathways using regional anesthesia and multimodality analgesia have decreased overall opioid use, but the additional benefits of robotic surgery in opioid reduction for pancreatoduodenectomy (PD) are unknown. We compared the inpatient opioid use between robotic PD and open PD.

Methods:

Patients undergoing open PD within a protocol evaluating pre-incisional regional anesthetic block bundles were compared to consecutively-treated patients undergoing robotic PD identified from a prospectively maintained single-institutional database. Clinical characteristics, operative outcomes, pain scores and inpatient oral morphine equivalent (OME) use were compared between patients treated with robotic or open PD. Patients with a history of continuous-release opioid dependence were excluded.

Results:

Of 114 total patients, 25 underwent robotic PD and 89 underwent open PD. Intraoperative opioid use was not different (p=0.87), nor were cumulative pain scores. Robotic PD patients used significantly fewer OMEs per day on postoperative days 1–4 (p=0.039), used fewer total OMEs during hospitalization (robotic: median=79, IQR 42.5–141; open: median=126, IQR 61.3–203.8; p=0.0036) and were discharged with fewer OMEs (robotic: median=0, IQR 0–43.8; open: median=25, IQR 0–75; p=0.009) despite a shorter length of stay (robotic: median=4, open: median=5, p=0.002).

Conclusions:

Robotic PD patients required fewer inpatient OMEs than open PD while maintaining similar pain scores. A higher percentage of robotic PD patients tapered off of opioids prior to discharge than open surgery counterparts treated with a standardized opioid reduction protocol despite a shorter length of stay. These results provide a rationale for choosing robotic PD when feasible to minimize opioid use.

Introduction

Robotic pancreatoduodenectomy (PD) utilization has increased significantly over the last decade as the use of robotic surgery becomes increasingly commonplace. Minimally invasive PD using laparoscopic approaches has not achieved popularity outside of a few high-volume centers owing to a steep technical learning curve and multiple studies demonstrating more severe postoperative complications or no benefits compared to open surgery^1,2. Laparoscopic and open PD were compared in the LEOPARD-2 trial, which was terminated early due to an observed high rate of complication-related death after laparoscopic PD³. In contrast, the robotic surgical platform offers potential advantages over laparoscopy including increased surgical dexterity and improved three-dimensional visualization. Robotic surgery is supplanting both open and laparoscopic approaches for complex foregut and colorectal surgery in selected patients. Robotic PD has been reported to be both safe and technically feasible when performed by experienced surgeons, although a careful approach is needed to safely initiate a robotic PD program at select high-volume centers⁴.

Our departmental risk-stratified pancreatectomy clinical pathways and enhanced recovery pathways have decreased length of stay, opioid use and morbidity associated with PD^5,6. With continued improvement in the peri-operative and postoperative care of PD patients, robotic PD represents a promising approach to further improve clinical care for selected patients. Purported advantages of robotic PD include decreased blood loss, shorter lengths of stay, and possibly higher lymph node yields^7,8. Despite these reported advantages, the data are conflicting regarding the benefit of robotic PD to postoperative outcomes, including decreased pain and quicker recovery. Optimal benchmarks for opioid requirements following PD are not established but it is generally accepted that the least amount of opioid use while maintaining patient mobility is ideal. A recent study suggested that minimally invasive pancreatectomy patients required increased opioids following surgery compared to their open-surgery counterparts⁹, counter to many reports of minimally invasive surgeries for other diseases¹⁰.

In our study, we sought to determine the impact of robotic PD on postoperative pain, perioperative opioid use, and discharge opioid prescriptions, compared to open PD in patients treated with a tightly controlled multimodal pain management care bundle.

Methods

Study Design and Patients

Clinical data was gathered from a prospectively collected cohort of PD patients treated at The University of Texas MD Anderson Cancer Center^11,12 from March 2019 through May 2021 (Figure 1). Two groups were identified: a robotic PD group which was comprised of all consecutively performed robotic PD cases over the study period, and an open PD which was comprised of the entire control arm of a trial evaluating the efficacy of regional block. Patient demographics, clinical characteristics, clinical course, inpatient opioid use and discharge opioid prescriptions were evaluated. Variables collected included patient characteristics at the time of surgery, including age, sex, race/ethnicity, body mass index and surgical indication. Operative and clinical factors examined included operative time, use of adjunct pain medications, utilization of regional blocks, length of stay and patient-reported pain scores on a scale of 0 – 10 with 10 being the most severe pain and 0 being no pain. Opioid amounts were converted to oral morphine equivalents (OMEs) per previously published opioid conversion ratios^13,14. Patients on extended-release opioids or prolonged (>6 months) preoperative immediate-release opioid use were excluded from the study. This study was approved by the MD Anderson Cancer Center Institutional Review Board (protocol PA17–0726). Adverse events (AEs) were documented using the ACCORDION system within a prospective surveillance program aimed at collecting AEs and grading them within 90 days following surgery^11,15,16. Grading of AEs and postoperative pancreatic fistulas (POPFs) was performed by faculty together with two advanced practice providers prospectively as previously described¹⁶. International consensus criteria were used in the grading of AEs and POPFs^15,17. Telephone calls 90 days after surgery ensured accurate data collection.

Figure 1: Overview of study design and patient cohort selection.

An opioid reduction initiative was initiated in 7/2018 which included provider education and standardization of postoperative opioids for all pancreatectomy cases. Patients were selected from a period from 3/2019–5/2021 after implementation of the opioid reduction initiative was well established. Patients undergoing open pancreatoduodenectomy were selected from a contemporary trial evaluating regional block in open surgery cases while patients undergoing robotic pancreatoduodenectomy were selected as consecutive cases over the study period.

Opioid Analgesia Pathway

All robotic PD patients underwent surgeon-placed laparoscopic-guided transverse abdominus plane (TAP) block while all open PD patients underwent anesthesiologist-placed ultrasound-guided 4-point quadratus lumborum plus subcostal block using liposomal bupivacaine mixed with bupivacaine. After completion of surgery, all patients were placed on a tightly monitored opioid pathway consisting of 3 non-opioid adjuncts: oral celecoxib at 100 mg twice per day, acetaminophen at 650 mg every 6 hours, and methocarbamol 500mg every 6 hours, unless contraindicated. If intravenous patient-controlled analgesia was deemed appropriate, the initial setting was 0.1 mg of hydromorphone (1 mg OME every 10 minutes) as needed by the demand button, with no continuous infusion, with bolus dosing of 0.2 mg every 2 hours if further needed, before escalation per clinical evaluation. Opioids were weaned as tolerated with specific postoperative day targets for complete transition to oral pain medications. At the time of discharge, the “5x-multiplier” was used, taking the patient’s overall inpatient OME used within the last 24 hours and multiplying by 5 to calculate their discharge opioid prescription¹⁸.

Statistical Analysis

Patient clinical demographics were compared between robotic and open PD as follows. Categorical variables were compared between groups using the chi-square or Fisher exact test. The F-test was used to determine the equality of variances prior to statistical comparisons. The Wilcoxon rank-sum test was used to compare nonparametric continuous variables. The two-sided Student t-test with Welch’s correction was used to compare parametric continuous variables. Missing variables were excluded from statistical analyses. Univariable and multivariable logistic regression models were fitted to examine the associations between variables and inpatient OME use, defining “high” OME use as the upper quartile amount (190 OMEs and greater). Factors with p values less than 0.20 in univariable analysis were included in the primary multivariable model. Age, sex, and surgical approach (open vs. robotic) were kept in the model because they were considered relevant variables. A 2-sided p of <0.05 was deemed statistically significant. Statistical analysis was performed using JMP Pro, version 15 (SAS Institute, Cary, NC, USA), and the Stata 14.1 software program (StataCorp, College Station, TX, USA).

Results

Patient Characteristics

A total of 114 patients were identified and included within the study with 25 undergoing robotic PD and 89 undergoing open PD (Table 1). There were no significant differences in patient characteristics between cohorts regarding age, sex, race/ethnicity, body mass index or disease. The open group was significantly more likely to have undergone neoadjuvant chemotherapy (p=0.001) and/or neoadjuvant radiotherapy (p=0.02). During the study period, one minimally invasive surgeon was primary or co-surgeon for all robotic PDs while seven different surgeons performed open PDs.

Table 1:

Patient demographics and clinical characteristics

	Robotic	Open	p value
Age, mean (range)	63 (21–79)	64 (32–91)	0.55
Sex, no. (%)			0.50
Male	12 (48.0%)	50 (56.2%)
Female	13 (52.0%)	39 (43.8%)
Race/ethnicity, no. (%)			0.10
White	17 (68.0%)	69 (77.5%)
Black	4 (16.0%)	4 (4.5%)
Hispanic	1 (4.0%)	11 (12.4%)
Asian	2 (8.0%)	3 (3.4%)
Other	1 (4.0%)	2 (2.2%)
ASA, no. (%)			0.13
2	1 (4.0%)	0
3	24 (96.0%)	87 (97.8%)
4	0 (0%)	2 (2.2%)
Body mass index, mean (range)	26.8 (16.5–37)	26.2 (17.7–36.9)	0.64
Pathology, no. (%)			0.07
PDAC	13 (52.0%)	64 (72.0%)
IPMN	6 (24.0%)	7 (7.9%)
PNET	2 (8.0%)	7 (7.9%)
SPPT	1 (4.0%)	0
Adenoma	2 (8.0%)	2 (2.2%)
Acinar cell carcinoma	0	2 (2.2%)
Ampullary adenocarcinoma	1 (4.0%)	0
Cholangiocarcinoma	0	3 (3.4%)
Other	0	5 (5.6%)
Tumor size (cm), mean (range)	2.6 (0.8–4.4)	2.8 (0.5–7.2)	0.19
Neoadjuvant chemotherapy, no. (%)	9 (36.0%)	63 (70.8%)	0.001
Neoadjuvant radiation, no. (%)	6 (24.0%)	43 (48.3%)	0.02

ASA, American Society of Anesthesia score; PDAC, pancreatic ductal adenocarcinoma; IPMN, intraductal papillary mucinous neoplasm; PNET, pancreatic neuroendocrine tumors; SPPT, solid pseudo-papillary tumor

Short-Term Operative Outcomes

In the robotic PD group, operative time was significantly longer (p=0.02), and estimated blood loss was significantly lower (p<0.001; Table 2). Median length of stay was 1 day shorter in the robotic PD group (4 vs. 5 days, p=0.002) with significantly more variability seen in the open cohort. Surgical complications were not significantly different between groups based on both International Study Group of Pancreatic Surgery grades and Accordion severity grading^15,17 (Table 2).

Table 2.

Operative outcomes

	Robotic n=25	Open n=89	p value
Operative time, median (range)	522.5 (400–625)	450 (279–772)	0.003
Estimated blood loss, mL, median (range)	100 (30–600)	200 (50–1500)	<0.001
Length of stay	4 (3–14)	5 (3–57)	0.002
Pain scores	2 (0–5)	2 (0–7)	0.98
Readmission, no. (%)	6 (24.0%)	20 (22.5%)	1.0
ISGPS Pancreatic fistula, no. (%)			0.71
None	21 (84.0%)	73 (82.0%)
Biochemical leak	2 (8.0%)	8 (9.0%)
Grade B	1 (4.0%)	7 (7.9%)
Grade C	1 (4.0%)	1 (1.1%)
Complications by Accordion Severity Grade 3 and greater	6 (24.0%)	17 (19.1%)	0.58

ISGPS, International Study Group of Pancreatic Surgery

Opioid Use

Intraoperative opioid use was not significantly different between the two groups (p=0.87, (Figure 2A) despite longer operative times in the robotic PD group. Robotic PD patients used fewer total OMEs during their hospitalization (robotic: median=79, interquartile range [IQR] 42.5–141; open: median=126, IQR 61.3–203.8; p=0.0036; Figure 2B) and were discharged with fewer OMEs (robotic: median=0, IQR 0–43.8; open: median=25, IQR 0–75; p=0.009; Figure 2C). In the RPD cohort, 56% of patients required no opioids within the last 24 hours prior to discharge compared to 43% of open PD patients. We then broke down the inpatient opioid use by postoperative day to identify when robotic PD patients began to use less opioids during admission and found that robotic PD patients were using less OME by postoperative day 1 and continued to have lower opioid use from that timepoint onward (Figure 2D). Despite robotic PD patients using less opioids, patient-reported pain scores were not significantly different between the two groups (robotic: median=2, open: median=2, p=0.98; Figure 2E), with evidence of higher variation in the open PD group.

Figure 2: Opioid use in robotic and open pancreatoduodenectomy cases.

a) Intraoperative opioid use was compared between open and robotic cases without significant differences. b) Total inpatient opioid use was significant lower in robotic pancreatoduodenectomy cases (robotic: median=79, interquartile range [IQR] 42.5–141; open: median=126, IQR 61.3–203.8; p=0.0036) and were c) discharged with fewer opioids (robotic: median=0, IQR 0–43.8; open: median=25, IQR 0–75; p=0.009). d) Inpatient opioid use was significantly less in the robotic pancreatoduodenectomy cohort from postoperative day 1 onward. e) Patient reported pain scores on a scale of 0–10 were not significantly different between the two groups. ns, non-significant, * represents p-value<0.05.

We then performed univariate and multivariate analyses to determine which factors correlated with OME use within the upper quartile. On univariate analysis, the robotic approach was associated with lower OME (odds ratio [OR] 0.2, 95% CI 0.04–0.091, p=0.037) while the presence of a pancreatic fistula was associated with higher OME (OR 3.94, 95% CI 1.44–10.85, p=0.008; Table 3). On multivariate analysis, these associations remained significant (robotic approach: OR 0.16, 95% CI 0.03–0.81, p=0.026; pancreatic fistula: OR 4.4, 95% CI 1.47–13.2, p=0.008; Table 3) as well as patients who were older (OR 0.96, 95% CI 0.92–1.00, p=0.036; Table 3).

Table 3.

Univariate and multivariate analysis to identify factors associated with upper-quartile OME use

	Univariable analysis				Multivariable analysis
Variable	Odds ratio	95% CI		p value	Odds ratio	95% CI		p value
Surgical approach, robotic vs. open	0.20	0.04	0.91	0.037	0.16	0.03	0.81	0.026
Age, older vs. younger	0.97	0.93	1.00	0.078	0.96	0.92	1.00	0.036
Sex, male vs. female	0.66	0.28	1.55	0.338	0.53	0.20	1.39	0.194
Body mass index	1.03	0.94	1.12	0.577
ISGPS, Pancreatic fistula	3.94	1.44	10.85	0.008	4.40	1.47	13.2	0.008
Complications by Accordion Severity Grade 3 and greater	1.29	0.31	5.34	0.729

ISGPS, International Study Group of Pancreatic Surgery

Discussion

In this study, we investigated our own experience with robotic PD and its impact on opioid use in a tightly monitored group of prospectively followed patients. We found that robotic PD was associated with decreased inpatient opioid use with similar postoperative pain scores. In both robotic PD and open PD groups, we found relatively low opioid usage due to several iterative changes we have made across our Department of Surgical Oncology for postoperative analgesia across disease sites^{5,13,14,18–26}. Lower inpatient opioid use then led to reduced discharge prescription volumes overall. On multivariate analysis, the robotic approach, absence of a pancreatic fistula and older age were associated with decreased OME. Consistent with previous studies, length of stay was 1 day shorter in the robotic PD group, at the cost of slightly longer operative times.

Opioid use disorder and illicit dissemination have remained a severe public health problem in the United States. Overdose deaths have continued to climb since 1999, with a steeper increase in deaths from 2013 onward and 500,000 deaths in the past two decades²⁷. While a major cause of the continued increase in opioid-related deaths is the proliferation of fentanyl and other synthetic opioids, individuals at particularly high risk for overdose are those with a history of opioid use prior to the expansion of fentanyl and are fentanyl naïve²⁸. This context is particularly relevant for prescribing physicians because while physicians rarely prescribe fentanyl postoperatively, the unintended escalation of opioid potency seen in illicit drugs is often what leads to overdose. Surgery is often the first exposure to opioids for many individuals²⁹. Lee et al. demonstrated that among patients who underwent cancer surgery, persistent opioid use past the postoperative period continued in 7% to 11% of patients who did not receive adjuvant chemotherapy and 15% to 21% of those who did receive chemotherapy. For patients looking for opioid sources outside of physician prescriptions, there is a significantly high risk of contamination with other dangerous drugs. Therefore, one of the goals of minimally invasive surgery is to minimize postoperative pain and, as much as possible, limit the need for opioids at time of discharge. With the use of non-opioid adjuncts and regional blocks, we were able to reduce median discharge OMEs to 25 within our open PD group and 0 within our robotic PD group, despite our robotic PD group leaving the hospital a day earlier. Comparatively, a recent study analyzing the opioid use following RPD demonstrated that within their cohort they had increased discharge opioid use and inpatient opioid use within their robotic cohort⁹. Within their discharge cohort, open PD received 30 oral morphine equivalents compared to 45 oral morphine equivalents in their RPD group. A trend towards higher opioid requirements were seen over the total inpatient period with RPD. The amount of opioids given over the hospitalization in this study is similar to other studies performed within the United States³⁰. In our study, 56% of our RPD patients required no opioids within the last 24 hours prior to discharge and were then discharged without an opioid prescription.

It is notable that despite a significant reduction in opioid use, the overall opioid utilization in both groups was relatively low. In August 2018, our institution unveiled an initiative to help curb unnecessary opioid use^14,23,24. This included provider education²³, intravenous patient-controlled analgesia setting guidelines⁵, and increased use of a non-opioid bundle with acetaminophen, celecoxib, and methocarbamol, and pre-incision regional nerve blocks. Our open PD cohort was taken from a comparison group in a contemporaneous randomized clinical trial of patients receiving a regional block for open surgery. The use of this cohort allowed for close monitoring and tightly controlled opioid dose escalation within the open group to help delineate the true impact of robotic surgery. We were able to perform detailed evaluation of opioid use within both groups and found that the opioid reduction was most notable within the first 24 hours after surgery. This result intuitively makes sense, as this period is when postoperative pain is usually the highest (depending on the success of the regional block) and when patients are tasked with rapid increases in their mobility. The reduction in opioid use continued for the remainder of the hospitalization, with a steady decrease in OME each postoperative day in both groups. It is also noted that pain scores between the two groups were similar. One may anticipate that general pain scores following minimally invasive surgery would be lower than open surgery. This may be the case without pain medication but as our clinical pathways are dependent on minimizing pain to facilitate ambulation, we are not able to determine relative pain scores at similar opioid dosing. Our use of pain medication is titrated to a point where patients are able to ambulate and perform tasks for daily living which typically approximately 3 on a scale of 10. The increased opioid requirements to reach a pain score of 3 is therefore an indirect measure of overall increased pain following open surgery relative to RPD. Studies of minimally invasive pancreatoduodenectomy have demonstrated similar trends on the need for postoperative analgesia^31–33.

There are several limitations to our study. First, because our study was not a randomized controlled trial, there were some differences identified between our open and robotic cohorts. The open cohort was taken from a contemporary control group evaluating regional blocks while our robotic PD group was comprised of consecutively performed robotic PD over the study period. The choice of open versus robotic approach was left up to surgeon discretion and therefore the robotic cases may represent less complex cases than the open group. These differences in cohorts is seen in the open group having a higher percentage of pancreatic ductal adenocarcinoma and a higher rate of receipt of neoadjuvant therapy. Another limitation inherent to our study design is surgeon differences between the two cohorts. The robotic PD at our institution were performed primarily by one surgeon (either as primary or co-surgeon) while the open cohort PDs were performed by 5 surgeons. Beyond surgeon differences, the robotic PD group’s postoperative analgesia pathway was not as strictly protocolized as the open PD group and may also contribute to the differences noted in narcotic use. Additionally, we do not know the exact date of complete opioid cessation for all patients included in the study. Finally, our open patients received pre-incisional anesthesia–placed ultrasound-guided blocks while robotic surgery patients had laparoscopically placed regional TAP blocks. While TAP blocks have been shown to reduce opioid use following robotic pancreatectomy³⁴, the effectiveness of surgeon- (laparoscopic) versus anesthesiologist- (ultrasound-guided percutaneous) placed regional blocks for pancreatic surgery have not been directly compared. Despite these limitations, we feel this study pragmatically suggests that robotic surgery can facilitate accelerated opioid tapering postoperatively.

Conclusions

Robotic PD was associated with less total inpatient opioid use, resulting in decreased opioid prescriptions at discharge. The impact appeared to be greatest in the first 24 hours after surgery, when patient mobility is most impacted by surgical approach. Robotic surgery did increase operative times but also was associated with a significantly shorter hospital length of stay in our cohort. These data suggest that a robotic surgical approach is beneficial in appropriately selected patients. However, randomized studies assessing cost and outcomes are needed to determine the true impact of robotic PD.