Association of Patient Controlled Analgesia and Total Inpatient Opioid Use after Pancreatectomy

Abstract

OBJECTIVES:

The initial settings on an intravenous patient-controlled analgesia (IV-PCA) pump can represent a significant source of postoperative opioid exposure. The primary aim of this study was to evaluate the impact of first day IV-PCA use on total inpatient opioid use after open pancreatectomy, before and after standardization of initial dosing.

METHODS:

Inpatient oral morphine equivalents (OME) were reviewed for pancreatectomy patients treated with IV-PCA at a single institution before and after (3/2016–8/2017 vs. 3/2019–11/2020) implementation of a standardized initial IV-PCA dosing regimen (initial limit 0.1mg hydromorphone, or 1mg OME, every 10 min as needed). IV-PCA OME in the first 24 hours and total inpatient OME were compared between cohorts.

RESULTS:

Of 220 total patients, 132 were in the pre-standardization historical cohort. First-24-hour IV-PCA use was different (pre-standardization median 95mg vs. post-standardization 15mg, p<0.001). Median total inpatient OME was different (p<0.001) between pre-standardization (525mg, IQR 239–951mg) and post-standardization patients (129mg, IQR 65–204mg) with 77% (median 373mg) of total inpatient OME contributed by IV-PCA in the pre-standardization and 56% (median 64mg) in the post-standardization cohorts. There were similar patient-reported pain scores between groups.

CONCLUSIONS:

Standardizing initial IV-PCA settings was associated with reduced first-24-hour opioid exposure, proportional and absolute total IV-PCA use, and total inpatient OME. Because of the contribution of an IV-PCA to total inpatient opioid exposure, purposeful reduction or omission of an IV-PCA are critical to perioperative opioid reduction strategies.

INTRODUCTION

Opioids remain a major societal burden despite numerous public health initiatives aimed at decreasing opioid overuse and medication diversion. In 2020, there were approximately 95,000 drug overdose deaths in the United States, up over 30% from the prior year.¹ Often, patients’ first exposure to opioids occurs after surgery with up to 15% going on to continued use well beyond the postoperative period.^{2, 3} This behavior is enabled through over-prescription which has led to excesses in disseminated opioids.^{3, 4} Prescribing practices have evolved in the wake of the opioid epidemic with new discharge prescription guidelines based on the surgical procedure performed or total inpatient opioid use.^5–9 In general, the focus of opioid research has been predominantly focused on prescribing practices, such as discharge opioid prescription heterogeneity, optimizing prescribing amounts and characterizing risk factors for continued use.^10–16 The move towards enhanced recovery and risk-stratified inpatient clinical pathways has decreased overall inpatient opioid consumption by limiting opioid use in the perioperative period by increasing the use of adjunct pain medications and regional anesthetic blocks.^17–19 While these efforts have been fruitful, enhanced recovery pathways and clinical pathways often do not have recommended initial dosing guidelines for opioids following complex surgical operations.

Intravenous patient-controlled analgesia (IV-PCA) pumps are typically used in large abdominal operations and can represent a significant amount of opioid use and initial opioid exposure during a hospitalization. While they are patient-controlled, the patients often do not have the knowledge of their dosage settings, the anticipated amounts to be used or a defined expectation for their postoperative pain. Despite our reliance on IV-PCAs for pain control, there are no studies looking at the contribution of IV-PCAs to total inpatient use of opioids in patients undergoing complex gastrointestinal surgery such as pancreatectomy and there exist no suggested guidelines for optimal IV-PCA dosing. The primary aim of this study was to determine the contribution of IV-PCA opioids to total inpatient opioid usage and to measure any change in opioid use after standardization of initial IV-PCA dosing in our care pathways.

MATERIALS AND METHODS

Study Design and Patients

This is retrospective cohort study comparing a historical cohort of pancreatectomy patients to our first cohort of patients treated with standardized IV-PCA settings, all at the University of Texas MD Anderson Cancer Center. All surgical outcomes were already included in our prospectively maintained pancreatic surgery database^{20, 21}. The first baseline cohort was from March 1, 2016, to August 30, 2017. These dates correspond to the rollout of our electronic health record and our initial ability to study our baseline opioid use after pancreatectomy through automated data extraction, followed by spreadsheet conversion to oral morphine equivalents (OME). This patient cohort overlapped with previous studies on discharge opioid use and departmental opioid use across 5 disease sites^{19, 22, 23}. Patients from August 31^st, 2017 – February 28^th, 2019 were not included in this study due to the iterative implementation phase of risk-stratified pancreatectomy clinical pathways during this time period. In the pre-standardization cohort, no boundaries were placed on postoperative IV-PCA initial dosing settings which were dosed at the discretion of the provider placing postoperative orders. In January 2019, our risk-stratified pancreatectomy clinical pathway^{20, 24, 25} incorporated low dose IV-PCA dosing. In March 2019, as part of a randomized trial on pre-incisional regional anesthetic block bundles in open surgery, it was standardized to an initial limit in the recovery room of 0.1mg hydromorphone (1mg OME every 10 minutes) as needed by demand button, with no basal settings, and bolus dosing of 0.2mg every 2 hours if further needed, before escalation per clinical evaluation. This education was distributed by printed pathway handouts and edited in our online order sets. The post-implementation cohort included open pancreatoduodenectomy patients from March 1, 2019, to November 30, 2020, chosen due their use as a contemporary control for surgical outcomes in our initial experience of minimally invasive pancreatoduodenectomy. Within the pre-standardization (PRE) and post-standardization (POST) guideline initiation cohort, postoperative pain medications (including the IV-PCA) were ordered by the primary surgical team. If patients had an epidural, management was performed by the acute pain service until the epidural was removed.

Variables collected included patient characteristics at time of surgery including age, sex, race, body mass index (BMI), and indication. Operative and clinical factors were also examined including the surgical extent, operative time, use of adjunct pain medications, utilization of pre-incisional regional blocks, length of stay, and patient-reported pain scores. The primary endpoint was first-24-hour IV-PCA OME. Secondary endpoints included total inpatient opioid use for the duration of the hospitalization and the proportion of the total which was attributable to the initial IV-PCA use. Opioid amounts were converted to oral morphine equivalents (OME) as per institutional conversion tables.¹⁹ This study was approved by the Institutional Review Board (protocol PA17–0726) with waiver of informed consent.

Statistical Analysis

Patient clinical demographics were compared using descriptive statistics. Comparison between categorical variables was performed using the χ2 or Fisher exact test. The F-test was used to determine the equality of variances prior to statistical comparisons. The students t-test was used to compare parametric continuous variables. The Mann-Whitney U was used to compare nonparametric continuous variables. Missing variables were excluded from statistical analyses. A 2-sided P < .05 was deemed statistically significant. OME data are represented as median with interquartile range (IQR) unless otherwise specified. Statistical analysis was performed using JMP Pro, version 15 (SAS Institute Inc, Cary, NC).

RESULTS

Patient Characteristics

Of the 220 evaluable patients in the 2 study periods, 132 were during the PRE period and 88 in the POST period. There were no significant differences in patient characteristics between cohorts regarding age, sex, race, BMI, disease or operative time (Table 1). The post-standardization group had a significantly lower length of stay compared to the pre-standardization group, due to the iterative improvements in our risk-stratified pancreatectomy clinical pathways. Surgical complications were not significantly different between groups based on both International Study Group of Pancreatic Surgery grades and ACCORDION severity grading^{26, 27}. IV-PCA was paired with intraoperative regional anesthetic block with liposomal bupivacaine and bupivacaine mixed in 94% of patients in the PRE group and 100% of patients in the POST group (p<0.001), representing a secular trend of eliminating epidurals in our usual care for pancreatectomy patients over the past 5 years. Of the 15 patients in the PRE group who did not receive a regional block, 7 received an epidural, 4 underwent laparoscopic distal pancreatectomies, and 4 had only IV-PCA (no block or epidural) for open pancreatoduodenectomy. The acute pain service managed the opioids in 15 of the 132 patients within the PRE group and none within the POST group.

Table 1.

Patient demographics and clinical characteristics

	Pre-Standardization N=132	Post-Standardization N=88	p value
Age, y, mean (range)	65 years (25–82)	66 years (32–91)	0.27
Sex	Female 59 (45%)	Female 38 (43%)	0.89
	Male 73 (55%)	Male 50 (57%)
BMI, mean (range)	26.24 (17.8–43.9)	25.85 (17.7–36.9)	0.16
Length of stay, median (range)	7 (3–51)	5 (3 – 57)	<0.001
Disease	PDAC 87 (65.9%)	PDAC 63 (71.6%)	0.59
	IPMN 9 (6.8%)	IPMN 7 (8.0%)
	PNET 18 (13.6%)	PNET 7 (8.0%)
	Other 18 (13.6%)	Other 11 (12.5%)
Surgery	Open pancreatoduodenectomy 98 (74%)	Open pancreatoduodenectomy 88 (100%)	<0.001
	Open distal pancreatectomy 24 (18%)
	Laparoscopic distal pancreatectomy 6 (4%)
	Central pancreatectomy 2 (2%)
	Total pancreatectomy 2 (2%)
OR Time	424 Minutes (149–913)	450 Minutes (279–493)	0.18
Pancreatic Fistula			p=0.37
None	95 (72.0%)	72 (81.8%)
A	15 (11.4%)	8 (9.1%)
B	19 (14.4%)	7 (8.0%)
C	3 (2.3%)	1 (1.1%)
ACCORDION Severity Grade			p=0.12
0	95 (72%)	72 (81.8%)
1	14 (10.6%)	8 (9.1%)
2	2 (1.5%)	4 (4.5%)
3	19 (14.4%)	4 (4.5%)
4	1 (0.8%)	0
5	0	0
6	1 (0.8%)	0

BMI, body-mass index, PDAC, pancreatic adenocarcinoma, IPMN, Intraductal papillary mucinous neoplasm, PNET, pancreatic neuroendocrine tumor, Pancreatic Fistula Grade based on International Study Group of Pancreatic Surgery grades²⁶. ACCORDION severity grading system²⁷.

Opioid Usage

Compliance with the IV-PCA guidelines in the POST cohort was 92% (81/88). First-24-hour IV-PCA use was significantly different with the PRE group having higher IV-PCA OME (PRE: median 95mg vs. POST: 15mg, p<0.0001, Figure 1A). Total IV-PCA use remained increased in the PRE group for the total inpatient stay with median total IV-PCA OME higher (median 373mg, IQR 163–748mg) than POST patients (64mg, IQR 13–146mg, p<0.001, Figure 1B). Total inpatient opioid use (IV and oral) was greater in the PRE cohort (PRE: median 525mg (IQR 239–951mg) vs. POST: 129mg (IQR 65–204mg), p<0.001, Figure 1C). In the PRE patients, IV-PCA total OME accounted for 77% of the total inpatient opioid use while in the POST group, IV-PCA dropped to 56% of the lesser total inpatient OME (Figure 1D).

Figure 1.

Intravenous patient-controlled analgesia (IV-PCA) contributions to inpatient opioid use. A) Median IV-PCA use in first 24 hours. B) Median total IV-PCA use per patient during the entire hospitalization. C) Total (IV and oral) inpatient opioid use. D) Total IV-PCA contribution to total inpatient opioid. Median boxplot data represented with Tukey whiskers (1.5 times the interquartile range). PCA, patient-controlled analgesia; OME, oral morphine equivalents; POD postoperative day

Subgroup Analysis of Open Pancreatoduodenectomy

In this retrospective cohort study, there existed baseline differences within the two groups regarding the extent of surgery. A subgroup analysis was performed where only open pancreatoduodenectomies were evaluated. The findings of increased IV-PCA first day OME, increased total IV-PCA OME and increased total inpatient OME in the pre-standardization group remained when limiting our study to only open pancreatoduodenectomy (Supplementary Table 1).

Pain Scores and Non-Opioids

Patient-reported pain scores were not different (statistically or clinically, all being in the 2–3 range out of 10) between the two groups in the first 4 postoperative days (Table 2), but the use of non-opioid adjunct medication evolved (Table 3) with greater use of anti-inflammatories and muscle relaxers in the POST group (p<0.001). This resulted in a different total number of adjunct non-opioid medications used (Table 3, p<0.001).

Table 2.

Pain score before and after standardization of intravenous patient-controlled analgesia initial settings

Time after surgery	Pre-Standardization (mean ± SD)	Post-Standardization (mean ± SD)	p value
POD 0–1	3.19 ± 1.76	2.87 ± 1.66	0.19
POD 1–2	2.89 ± 1.45	3.21 ± 1.31	0.09
POD 2–3	2.49 ± 1.40	2.52 ± 1.31	0.87
POD 3–4	2.04 ± 1.40	2.32 ± 1.47	0.16

POD, postoperative day; SD, standard deviation

Table 3.

Use of adjunct non-opioid pain medications

Medication	Pre-Standardization N = 132	Post-Standardization N = 88	p value
Acetaminophen	131 (99%)	87 (99%)	p>0.99
NSAIDs	81 (61%)	62 (70%)	p=0.19
Methocarbamol	68 (52%)	80 (91%)	p<0.001
Gabapentin	12 (9%)	7 (8%)	p=0.81
Number of adjunct medications (mean ± SD)	2.81 ± 1.33	3.66 ± 0.71	p<0.001

DISCUSSION

In this study of IV-PCA use in pancreatectomy patients, we found that IV-PCA use accounted for the vast majority of opioid use in the index hospitalization. Simply setting initial guidelines on the IV-PCA dosing for surgical providers placing electronic orders decreased the first-24hr spike in IV-PCA contribution to total inpatient opioid use, total IV-PCA OME, the proportion of the total inpatient OME that was from the IV-PCA, and the absolute total inpatient opioid use. Considering the operative and management complexity of pancreatectomies, this no-cost change is both effective and generalizable across any major abdominal surgical procedure.

While the IV-PCA has become a staple of post-surgical care, little attention is given to standardized dosing guidelines for specific procedures. The IV-PCA initially gained popularity in the 1970s when Marks and Sachar published on the large proportion of hospitalized patients that were undertreated for their pain.²⁸ At the time, these findings were very surprising and led providers to reevaluate our pain medication practices. By 1976, the first commercially available PCA was developed which allowed for more patient autonomy and limited the burden on nursing to provide pain medication.²⁹ It also allowed for lower doses to be given more often, so patients could achieve a more steady state of relief rather than larger IV boluses. While these devices have revolutionized the care of the post-operative patient and reduced patient pain, they are still reliant on physician dosing. Multiple studies have shown that provider perceptions of perioperative pain and opioid needs vary widely.^30–32 Our own institution has demonstrated significant variation in the expected opioid requirements following oncologic surgery among provider types (fellows, attendings and advanced practice providers) and that the practice variation can be mitigated with educational intiatives.³¹ Providers in the United States have also consistently been shown to anticipated greater opioid needs compared to their international counterparts.^{30, 32–34} U.S. providers have been shown to expect higher pain scores at discharge, estimate more days of opioid use post-operatively and recommend greater opioid prescriptions following oncologic surgery.³⁰ Education initiatives and prescribing guidelines reduce variation, and as evidenced by our study, IV-PCA guidelines can be incorporated into risk-stratified pathways to reduce unnecessary opioid use. This is particularly imperative at academic centers where there is inherent monthly and yearly turnover in surgical trainees, resulting in more practice variation.

In conjunction with the decreased use of IV-PCA over this time period, we saw an increase in overall non-opioid adjunct use, specifically methocarbamol. Methocarbamol is a muscle relaxant used primarily for painful musculoskeletal spasms with a poorly understood mechanism thought to act centrally on the nervous system.^{35, 36} Methocarbamol has been shown to be a useful adjunct for postoperative patients in a variety of surgeries and after trauma.^37–39 While methocarbamol was the only specific medication seen to be significantly different between the PRE and POST groups, the overall number of adjunct medications was higher in the POST group. These findings demonstrate that multi-modal pain regimens decrease opioid use as previously shown⁴⁰. Wei et al. showed that implementation of a multimodality pain regimen resulted in significant reductions of inpatient opioid use and pain scores in trauma patients. While open pancreatectomy patients differ significantly to the patients in their study, we find here that the addition of numerous pain medications acting through multiple mechanisms can limit opioid use. For our patients, initiation of methocarbamol is routinely started immediately postoperatively if there is no contraindication and continued through the hospitalization. It is available in both IV and oral forms which makes it a practical option in patients following major gastrointestinal surgery. Additionally, unless contraindicated, we start oral celecoxib 100mg twice per day and acetaminophen 650mg every 6 hours within the first 24 hours.

While IV-PCA use can drastically increase total opioid use for the hospitalization, they remain a traditional staple of the post-operative recovery period as adequate patient pain control is vital to early mobility and recovery. In a randomized clinical trial, Klotz et al. evaluated gastrointestinal complications after pancreatoduodenectomy in patients receiving either an epidural or IV-PCA.⁴¹ Among the 248 study participants, there were no significant differences in gastrointestinal complications or any surgical complications between the two groups. IV-PCA was associated with a lower use of intraoperative vasopressors and less weight gain by post-operative day 4 while the epidural group was found to have a 18.5% failure rate. As enhanced recovery pathways have become the standard, regional blocks have gained popularity due to their ease of placement relative to the epidural and decreased likelihood of hypotension. Regional blocks are associated with decreased length of stay when compared to epidurals^{42, 43}. Our institution preference has shifted to regional blocks with low dose IV-PCA as a result of these studies and our own experience. The next iterative goal would be to eliminate the use of IV-PCAs or to truly limit them to 24 hours of use. If trauma patients can be managed without IV-PCAs, one would imagine that elective surgery patients could be as well^{40, 44}. Perhaps the putative limitation is our own traditional concern about using the stomach for pills immediately after a pancreatoduodenectomy. But with our reduced reliance on nasogastric tubes²⁵, perhaps IV-PCA omission is possible selectively when bundled with a sufficient oral multimodal non-opioid bundle.

Our study has several limitations including the retrospective nature of the cohorts chosen, and the data limited to a single institution experience. We did not control for dosing settings in the PRE group and therefore the actual dosing within that group was quite variable with typical dosing between 0.1–0.4 mg of hydromorphone q6-15min with variable lockout limits. Additionally, nursing boluses of 0.5–2.0 mg were available for breakthrough. While this led to significant variability in opioid use within the PRE group, this is likely a reflection of real-world conditions where guidelines are nonexistent. The POST group was treated during a period when there was more team emphasis on minimizing overall opioid use and order-set-guided reduced variation in initial opioid orders (outstanding 92% compliance), and this likely had an inherent impact based on the data, especially total opioid use after the first 24 hours as we emphasized accelerated weaning as one of four pillars of opioid reduction^{5, 6}. Additionally, our two groups had different distributions of the operations performed, with the POST group undergoing larger operations (only open pancreatoduodenectomies). Despite the POST group undergoing larger surgeries, the finding of decreased opioid use remained and was seen in subset analysis of only open pancreatoduodenectomies. As with any single-institution retrospective cohort study, the most salient criticism is always the external validity, or generalizability, of the findings. Most surgeons would accept the argument that pancreatectomies remain one of the most complex abdominal operations to successfully perform and manage postoperatively. Thus, if our straightforward IV-PCA changes can be accomplished with no reflexive changes in patient-reported pain scores, then seemingly this can be done for all types of abdominal surgery in any institution looking to reduce opioid use. Despite these limitations, we were able to show a large difference in the reduced utilization of IV-PCA with zero cost and high compliance.

Conclusions

Standardizing initial IV-PCA settings was associated with reduced first-24-hour opioid exposure, proportional and absolute total IV-PCA use, and total inpatient OME. This was accomplished without reflexive changes in patient-reported pain scores. Because of the contribution of IV-PCAs to total inpatient opioid exposure, their purposeful reduction or omission are critical to perioperative opioid reduction strategies. This no-cost intervention is reproducible and when coupled with other pillars of opioid reduction can have substantial positive impact.