Continuation versus Interruption of Buprenorphine/Naloxone in Adult Veterans Undergoing Surgery: Examination of Postoperative Pain and Opioid Utilization in a National Retrospective Cohort Study

Abstract

Background:

Prescription rates for buprenorphine in opioid use disorder are increasing, and recent guidelines recommend its continuation during and after surgery; however, evidence from clinical outcome studies is limited. We tested the hypotheses that 1) perioperative continuation of buprenorphine does not result in higher pain scores and 2) that this approach does not result in higher supplemental postoperative opioid requirements.

Methods:

The Veterans Affairs Corporate Data Warehouse was queried for patients who underwent surgery while being prescribed buprenorphine/naloxone for opioid use disorder between 2010 and 2020. Analysis of the prescription record was used to infer buprenorphine management, and a 3:1 matched control set of patients without buprenorphine prescriptions was generated. We examined patients who continued buprenorphine, patients who had buprenorphine interrupted, and control patients. The primary outcome was time-weighted average postoperative pain scores from inpatient and outpatient sources within 72 hours of surgery. The secondary outcome was postoperative average daily morphine equivalent opioid requirements within two weeks of surgery.

Results:

A total of 1,881 surgical procedures in 1,673 patients taking buprenorphine for opioid use disorder were included; these procedures were matched to 5,748 control patients (5,775 procedures) without a buprenorphine prescription. Among the 1,881 procedures, 1,186 (63%) continued buprenorphine through the perioperative period while 695 (37%) interrupted buprenorphine. Pain scores were clinically similar for all three groups (4.1 ± 1.9 control [n = 3284], 4.9 ± 2.0 continued buprenorphine [n = 662], and 5.5 ± 1.7 interrupted buprenorphine [n = 419]; P < 0.001). Patients who continued buprenorphine did not require significantly more supplemental opioids as compared to controls (39.7 mg morphine equivalents/day ± 1.9 versus 36.5 ± 0.7, P = 0.23), and patients who interrupted buprenorphine received more supplemental opioids than those who continued it (74.2 ± 4.5 mg morphine equivalents/day versus 39.7 ± 1.9, respectively; P < 0.001).

Conclusions:

Continuation of buprenorphine is not associated with higher average pain scores or postoperative opioid requirements, supporting recently published guidelines.

Introduction

Opioid use disorder is a chronic disease affecting approximately 3 million Americans,¹ and opioid-related overdose is one of the leading causes of preventable mortality for Americans.² Many patients with opioid use disorder also suffer from pain. Buprenorphine is available in many forms (Supplemental Table 1) and was first developed in the 1970s as an analgesic that was subsequently found to be an effective medical treatment for opioid use disorder because it not only decreases opioid withdrawal and craving but may also decrease opioid abuse and mortality.^3–6

Medical treatment of opioid use disorder, including methadone and buprenorphine as opioid treatment options,^7,8 has been associated with improved outcomes and reduced mortality. In the United States, any Drug Enforcement Agency-licensed practitioner can now prescribe buprenorphine without regulatory limitations, which will likely result in more patients with opioid use disorder presenting for surgery while prescribed the medication. Current guidelines and expert opinions recommend continuing perioperative buprenorphine. Still, there have been concerns that high mu-opioid receptor binding, high receptor affinity, and partial antagonism of the receptor could complicate postoperative pain control. Evidence supporting continuation is derived from case studies, small case series, and retrospective observational studies. Case reports describe successful management of postoperative pain when buprenorphine was continued perioperatively,^9,10 and case series and retrospective studies have reported adequate postoperative pain control in patients on buprenorphine or methadone.^11–13

However, other case reports demonstrated that some patients do not achieve suitable analgesia while on buprenorphine.^14–17 Despite some differing conclusions for case reports, systematic reviews concluded that there was no evidence to support routine discontinuation of perioperative buprenorphine; however, high-quality evidence was limited.¹⁸ While the unique pharmacologic properties of buprenorphine could interfere with acute pain treatment, adequate pain control appears possible using full opioid agonists, and there is little evidence that interrupting therapy results in improved postoperative analgesia and may result in high postoperative dosing requirements.¹⁹

Recently published guidelines recommended against stopping buprenorphine therapy before surgery and that interruption may be detrimental to patients.^20,21 The American Society of Regional Anesthesia Guidelines recommended that buprenorphine should be continued in the perioperative period and that it can be initiated in patients with untreated opioid use disorder and acute postsurgical pain to decrease the risk of illicit opioid abuse and death from overdose after surgery.²⁰ There remains concern over the ability to manage acute pain with full opioid agonists, particularly for procedures with more postoperative pain, while administering buprenorphine.

We used a national Veterans Administration clinical registry and database to test the hypothesis that continuing buprenorphine in the perioperative period is not associated with higher postoperative pain or postoperative opioid prescription requirements. We also explored the relationship of multiple variables with postoperative pain outcomes, including patient factors, hospital facility complexity rating, surgical year, type of surgery (classified by the postoperative Pain Severity Scale), preoperative opioid use, and buprenorphine dose.

Methods

Data

The patient cohort was selected from the Veterans Affairs (VA) Corporate Data Warehouse, including any patient who received an outpatient opioid prescription (defined as Drug Class containing “opioid analgesic”) from January 2010 to December 2020. The Veterans Affairs Surgical Quality Improvement Program (VASQIP) data table was used to identify surgical procedures, the outpatient pharmacy tables and the inpatient Barcode Medication Administration tables were used to identify opioid prescriptions, and International Classification of Disease (ICD) codes were collected from inpatient and outpatient encounters to determine concurrent comorbid diagnoses. The VA Western New York Healthcare System Institutional Review Board approved this study, and a Health Insurance Portability and Accountability Act (HIPAA) consent waiver was granted. This was a preliminary observational study using data from the VA corporate data warehouse and surgical quality improvement database, and the data analysis and statistical plan were determined after the data were accessed.

Database queries were run on Microsoft SQL Management Studio (version 18.9.1) in the VA Informatics and Computing Infrastructure (VINCI) workspace. The results were imported into R-Studio (version 1.4.1717) for data cleaning and analysis.

Study Population

The database was queried for patients who received an outpatient prescription for buprenorphine/naloxone within 35 days of the operation date and received a diagnosis code for opioid use disorder within 1 year of the date of surgery (see Figure 1 for the flow chart diagram). The 35-day time window was chosen to include patients who could have used buprenorphine until 5 days before surgery, assuming a 30-day supply. The dose of preoperative buprenorphine prescribed nearest to the surgery date was collected for all case patients.

Figure 1:

Cohort Flow Diagram. Patients with an entry in the surgical outcome database with a prescription for buprenorphine/naloxone within 30 days and a diagnosis of opioid use disorder within 1 year were included. Patients were excluded if they had missing buprenorphine prescription information, had ASA class 5, Emergency modifier codes, or repeated surgery within 30 days.

Patients were included in the analysis if they had an American Society of Anesthesiologists (ASA) Physical Status Classification 1 through 4 and underwent elective surgery. To avoid analyzing complications related to very complex or emergency surgical situations, patients were excluded if they had an ASA physical status classification of 5 or underwent emergency surgery (“E” modifier code). A total of 2,064 cases in patients prescribed buprenorphine for opioid use disorder were initially identified (with 150 patients excluded for ASA Classification 5 or “E” modifier use).

Repeated surgical cases within 30 days of one another were removed (n = 28) to exclude possible surgical complications, resulting in 2,036 unique surgical procedure entries in the database. Detailed buprenorphine prescription information was missing in 155 patients, and those patients were excluded from the analysis, leaving 1881 surgical cases in 1673 individual patients.

A 3:1 control dataset was constructed from the surgical outcome database. Patients were excluded from the control cohort if they received buprenorphine of any type 90 days before surgery, but controls were not screened for preoperative full opioid agonist use. The controls were matched on age (± 5 years), gender, race, ASA classification, smoking status, postoperative Pain Severity Scale²² (see Feature Set Data below), and year of surgery (± 1 year). Patients with a repeat entry in the surgical outcome database within 30 days and patients with a diagnosis code for opioid use disorder within one year surrounding surgery were excluded from the control cohort, yielding 1,074,383 unique surgical cases. A total of 5,775 matched control patients were randomly selected and included in the matched-control set.

Exposure

The primary exposure variable was interruption of perioperative buprenorphine. Buprenorphine continuation or interruption was inferred with three sequential steps. First, the inpatient medication administration record was searched for administration of buprenorphine in the 36 hours after surgery, and patients who received buprenorphine were assigned to the continued buprenorphine cohort, and patients who were admitted for ≥ 36 hours and did not receive buprenorphine were assigned to the interrupted cohort. Next, the outpatient prescription instructions were searched for buprenorphine hold instructions using regular expression string pattern matching. Patients were assigned to the interrupted buprenorphine cohort when an instruction to hold the medication was found in the outpatient prescription record. Lastly, if a patient was not assigned to a continued or interrupted category by the abovenamed methods, the timeliness of the postoperative prescription of buprenorphine was calculated based on the outpatient prescription records by comparing the postoperative dispensed date to the expected dispensing date from the preoperative prescription and the number of days supplied. The length of hospital stay was incorporated into the expected postoperative due date for patients who were admitted to the hospital. Patients who received a postoperative prescription 4 or more days later than expected (after correcting for hospital length of stay) were assigned to the interrupted buprenorphine cohort. All remaining patients were assigned to the continued buprenorphine cohort.

The preoperative daily dose of buprenorphine closest to surgery was also calculated for all patients in the case group by calculating the maximum daily dose in milligrams from the preoperative prescription closest to the surgical date.

Outcomes

The primary outcome was time-weighted average postoperative pain recorded in the electronic health record. Pain Score data was collected from the Vital Sign data table, extracting entries with Vital Sign Type equal to “pain” and a numeric value between 0 and 10 (Numeric Rating Scale) within 72 hours of surgery. The data table includes information from clinical documentation in inpatient and outpatient settings but does not typically include information from the Intensive Care Unit or Post Anesthesia Recovery Unit because of frequent documentation and different documentation procedures (handwritten notes or separate documentation software, depending on the facility). Maximum 24-hour pain scores were recorded for each patient. A time-weighted average pain was calculated for the 72-hour postoperative time window for any patient having three or more pain score entries. The analysis did not include patients with fewer than three pain scores recorded. The time-weighted pain average is calculated by multiplying the sum of the average value of adjacent pain score observations by the time difference between the observations and then dividing the sum of weighted averages by the duration of the observation time window.²³ A difference of greater than 1 point in the numeric rating scale for pain was considered clinically relevant.²⁴

The secondary outcome was postoperative opioid requirements. We included orders for all opioids in the inpatient or outpatient pharmacy records within 30 days of surgery. Opioid medication orders were identified by searching for the drug class equal to “opioid analgesic”. Medication administration data was extracted from the inpatient pharmacy table, and prescription data was extracted from the outpatient pharmacy orders. Postoperative opioid prescriptions were extracted for patients across all three groups, including prescriptions for codeine, morphine, hydrocodone, hydromorphone, oxycodone, tramadol, fentanyl, methadone, and buprenorphine. Opioid doses were converted to milligrams of morphine equivalency (MME) as described below, and the average daily dose was calculated for the first 2 postoperative weeks (postoperative days 0 to 13) for each patient. The mean of the daily average opioid dosage and standard deviation were calculated for each cohort for the two-week postoperative time period.

Doses of opioids were standardized to milligrams of morphine equivalency using standard methodology; however, the Centers for Disease Control (CDC) no longer lists a morphine equivalency for buprenorphine because of its unique properties and favorable risk profile, so the previous conversion of 30:1 was used for part of the postoperative opioid analysis.²⁵ Our analysis of postoperative opioid utilization included both total postoperative opioid dosing that included buprenorphine and supplemental opioid dosing that examined the use of full opioid agonists and excluded buprenorphine.

A daily total opioid dose, expressed in milligrams of morphine equivalency, was calculated for each patient by combining data from both inpatient and outpatient sources. For outpatient prescriptions, the maximum daily dose was calculated and carried forward for each day the prescription was active. If there were overlapping outpatient prescriptions, their doses were summed. Inpatient medication administration events were also summed for each day with active prescriptions. Finally, the daily opioid doses from both inpatient and outpatient settings were combined to determine the total daily opioid dose for each patient.

Feature Set Data

We included the following covariates to control for possible confounding factors: surgical Pain Severity Scale (see below), year of surgery, VA facility complexity rating, postoperative day, preoperative buprenorphine dose, cohort (control, buprenorphine continued, and buprenorphine interrupted), sex, race, smoking status, and ASA physical status classification. VA facilities are assigned a complexity code based on the availability of subspecialty care, research, and teaching, with complexity 1A representing the highest complexity rating and 3 representing the lowest.

Each surgical procedure code was mapped to a Pain Severity Scale (PSS) group to account for the varying degrees of postoperative pain across the wide range of surgical procedures.²² In brief, the Pain Severity Scale was previously developed using k-means clustering to group surgical procedures found in the VA surgical outcome data table based on observed pain reports (maximum 24-hour pain and 72-hour average pain) and postoperative outpatient opioid prescriptions. The scale is a 5-level ordinal variable, with increasing scores indicating increasing postoperative pain or opioid consumption. Surgical procedures were mapped to a Pain Severity Scale value to partially control for variability in postoperative pain across a wide range of procedures. The buprenorphine dose was included as a continuous variable without any data transformation and modeled using a spline curve as described below.

Statistical Analyses

We did not perform a sample size or effect size calculation a priori. Our objective was to examine the entirety of the patient cohort from the VA surgical outcome database that underwent surgery while on buprenorphine for opioid use disorder. Initial data exploration of the database informed the subsequent structuring of outcomes and statistical methods.

The demographic variables were compared using two-sided t-tests for continuous variables and chi-squared tests for categorical variables. Bonferroni correction was applied for instances of multiple testing. The daily buprenorphine dose closest to surgery was collected for the groups that continued and interrupted the medication and significance testing was performed using Analysis of Variance (ANOVA).

We tested the primary hypothesis that continuing buprenorphine was not associated with higher pain scores by comparing the time-weighted average pain scores between the three cohorts. Significance testing was performed with a two-sided t-test and Bonferroni correction. We compared the pain outcomes for high and low-dose buprenorphine in the two case groups using Analysis of Variance (ANOVA).

We tested the secondary hypothesis that continuing buprenorphine was not associated with higher supplemental postoperative opioid requirements by comparing the average daily dose of postoperative opioids after converting the doses to morphine-equivalent doses as described above and excluding buprenorphine from the aggregate data. Total opioid administration was also calculated, including buprenorphine in the aggregate analysis. Opioid dosage data displayed a rightward skew due to high-dose outliers; thus, we calculated the significance with both mean opioid dose and a logarithmic transformation of the opioid dose; however, the significance estimates were not different using the two methods.

A predetermined sensitivity analysis was performed to examine the associations of covariate data with the primary and secondary outcomes. Analysis of Covariance (ANCOVA) was used to investigate the impact of covariates on the postoperative pain outcome, including the association of the surgical Pain Severity Scale, the facility’s complexity rating, the ASA classification, the buprenorphine dose category (high [> 16 mg/day] versus low [≤ 16 mg/day]), buprenorphine management (continued versus interrupted), and preoperative opioid use in the control patients (see post hoc sensitivity analysis below). A generalized additive model (GAM) was used to examine the association of independent variables on the daily opioid dose over the four weeks following surgery using the mgcv library in R (Mixed GAM Computation Vehicle with Automatic Smoothness Estimation).²⁶ Independent categorical variables included the surgical Pain Severity Scale, year of surgery, VA facility complexity rating, postoperative day (POD), preoperative buprenorphine dose, cohort (control, buprenorphine continued, and buprenorphine interrupted), sex, smoking status, ASA physical status classification, and preoperative use of opioids in the control group. The continuous variables (postoperative day [POD] and buprenorphine dose) were modeled using a penalized, thin plate regression spline provided by the mgcv library in R without applying any data transformation.²⁷

Post hoc sensitivity analysis was performed to evaluate the impact of missing pain data and the threshold for determining interrupted buprenorphine. The impact of missing pain data was assessed by imputing the missing value from the average pain score of patients who underwent the same surgical procedure. The preoperative opioid exposure was collected for the control patients and divided into three groups: no preoperative opioid exposure, > 40 mg of morphine equivalency averaged 60 days before surgery, and ≥ 40 mg of morphine equivalency. This threshold was chosen because it reflects a patient taking 10 mg of hydrocodone or oxycodone three to four times per day, reflecting consistent daily consumption of opioids. This categorical data was included in the sensitivity analyses described above.

Results

Demographics and Comorbidities

The patient demographics for the case and the control group are presented in Table 1. In total, data from 7656 patients were analyzed. Of those, 1881 cases (buprenorphine case group) received buprenorphine for opioid use disorder preoperatively, and 5775 were case-matched controls without buprenorphine prescriptions within 90 days prior to surgery. Patient characteristics were similar in age, gender, race, ASA Classification, and smoking status. The mean age was 53.3 years (standard deviation 12.5) and 54.1 (12.7) for the case and control populations. The population was predominantly white males, typical of a VA patient population. There were no statistically significant differences in demographic characteristics between the case and control cohorts (see Table 1). We included patients who had more than one surgical procedure during the study period (232 of 1737 patients in the buprenorphine group [13.4%] and 26 of 5748 patients in the control group [0.5%], and the average time between observations was greater than 600 days.

Table 1:

Patient demographics were similar between the groups, with the distribution of demographics typical of a VA population. The other race categories include declined, unknown, American Indian, and Native Hawaiian/Pacific Islander.

	Case (n = 1881)	Control (n = 5775)	Standardized Mean Difference
Age
mean (standard deviation)	53.3 (12.6)	54.1 (12.7)	−0.063
ASA Classification
2	517 (27.5%)	1582 (27.4%)
3	1271 (67.6%)	3903 (67.6%)
4	93 (4.9%)	290 (5.0%)	0.002
Sex
Male	1734 (92.2%)	5328 (92.3%)
Female	147 (7.8%)	447 (7.7%)	−0.003
Smoking Status
Smoker	1183 (62.9%)	3624 (62.8%)	0.003
Non-Smoker	698 (37.1%)	2151 (37.2%)
Race
White	1510 (80.3%)	4617 (79.9%)	0.008
Black	261 (13.9%)	829 (14.4%)	−0.014
Other	110 (5.8%)	329 (5.7%)	0.006

Delayed or Interrupted Postoperative Buprenorphine Prescription

Among 1,881 surgical procedures in patients with preoperative buprenorphine therapy, 1,186 (63%) continued the medication through the perioperative period and 695 (37%) interrupted buprenorphine therapy. The median preoperative buprenorphine dose of the buprenorphine group was 16 mg/day. Those who continued buprenorphine after surgery had a slightly higher average preoperative dose of 16.0 ± 7.5 mg compared to those who had interrupted buprenorphine at 14.7 ± 7.8 mg (P < .001), but this small difference is not clinically significant (Table 2). In addition to having a higher average dose, the group that continued buprenorphine had fewer patients on lower doses (8 mg/day or less) and more patients on higher doses (greater than 16 to 24 mg/day).

Table 2:

Distribution of preoperative buprenorphine daily dose for patients with opioid use disorder: The continued and interrupted buprenorphine cohorts had similar distributions of doses, but there was a slight incidence of higher buprenorphine daily dose in the group who had the medication continued during surgery. A chi-squared test was used to compare the distribution of the buprenorphine daily doses between the continued and interrupted cohorts (P < 0.001).

Buprenorphine Daily Dose (mg)	Continued	Interrupted
0.5 to 8	234 (25.9%)	303 (33.0%)
9 to 16	366 (40.6%)	363 (39.6%)
17 to 24	249 (27.6%)	196 (21.4%)
25 to 40	53 (5.9%)	55 (6.0%)

Continuation of perioperative buprenorphine administration was defined using a combination of inpatient buprenorphine administration, outpatient prescription instructions, and the timeliness of the first postoperative prescription (see Methods). Over the ten-year study period, 695 (36.9%) patients interrupted buprenorphine, and 1186 patients (63.1%) continued it. The percentage of patients who interrupted buprenorphine varied between 49.6% (2012) and 19.0% (2020), suggesting a changing approach to perioperative buprenorphine management over time (Table 3).

Table 3:

The incidence of interrupting buprenorphine for surgery was lower than previously published rates, possibly representing some patients misclassified as having continued buprenorphine. The rate of interruption declined in the later years of the study period.

Year	Continued	Interrupted
2010	39 (57.4%)	29 (42.6%)
2011	52 (51.0%)	50 (49.0%)
2012	61 (50.4%)	60 (49.6%)
2013	74 (56.5%)	57 (43.5%)
2014	87 (58.4%)	62 (41.6%)
2015	86 (54.4%)	72 (45.6%)
2016	108 (60.0%)	72 (40.0%)
2017	133 (65.8%)	69 (34.2%)
2018	172 (66.4%)	87 (33.6%)
2019	187 (66.8%)	93 (33.2%)
2020	187 (81.0%)	44 (19.0%)

Postoperative Pain Scores

Time-weighted pain scores were calculated for the 72 hours following surgery from pain intensity scores recorded in the vital sign data table, which includes inpatient and outpatient observations but does not typically include scores from the Post Anesthesia Recovery Unit or the Intensive Care Unit. Clinically small but statistically significant differences were observed between the groups’ average pain scores. The time-weighted average pain and standard deviation was 4.1 ± 1.9 for the control group, 4.9 ± 2.0 for the group that continued buprenorphine, and 5.5 ± 1.7 for the group that interrupted buprenorphine. The differences between the groups were statistically significant (P < 0.001) for all comparisons after applying the Bonferroni correction (Figure 2A). The difference in average pain scores was less than one point (the clinically relevant threshold) between the group that continued buprenorphine and the two other groups, while the difference between the group that interrupted buprenorphine and the control group pain scores was 1.4 points, with the group that interrupted buprenorphine having an average pain that falls within the moderate pain classification.

Figure 2:

A) Time-Weighted Average Postoperative Pain in the 72 Hours Following Surgery. Small differences were observed in the average pain scores between the groups (P < 0.001 for pairwise comparisons after Bonferroni correction), but the differences were not clinically relevant. B) The mean of the time-weighted average postoperative pain was similar for both buprenorphine groups (continued and interrupted) at both low (<= 16 mg/day) and high (> 16 mg/day) doses. The difference between the average postoperative pain was within 1 point on the numeric rating scale, but the small differences were statistically significant (P-value = 0.008; Analysis of Variance).

The buprenorphine cohort was divided into low doses (≤ 16 mg/day) or high doses (> 16 mg/day) based on the preoperative prescription (Figure 2B). The average pain was 4.8 ± 2.0 and 5.0 ± 2.0 for low and high doses in the cohort that continued buprenorphine, and it was 5.4 ± 1.7 and 5.9 ± 1.6 for the group that interrupted buprenorphine. These differences were not clinically or statistically different (P = 0.132). We performed a sensitivity analysis using analysis of covariance (ANCOVA) to ensure that other factors did not influence our results (Supplemental Table 2). These factors included the invasiveness of surgery (Pain Severity Scale), complexity rating of the facility, ASA classification, buprenorphine dose category, and buprenorphine management (continued versus interrupted). The facility complexity rating (P = .197) and the high buprenorphine dose (P = .091) did not have a significant association with the primary outcome. However, we found that buprenorphine management (P < 0.001), smoking (P = .006), and surgical Pain Severity Scale rating (P < 0.001) had statistically significant association with the measured differences in postoperative pain reports (Supplemental Digital Content).

Postoperative pain data were missing for many less invasive procedures (Pain Severity Scale Class 1 and 2; 76 and 79%, respectively), with less missing data encountered in intermediate procedures (Pain Severity Scale Class 3 and 4; 30% and 22%), and infrequent missing data in high-pain postoperative pain severity cases (8%). Missing data was more commonly encountered in outpatient procedures (70.5%) than when patients were admitted to the hospital (7.1%). Missing data was evenly distributed among the various patient cohorts (control = 43.1%, continued buprenorphine = 44.2%, and interrupted buprenorphine = 39.7%; P = 0.15). Post-hoc sensitivity analysis revealed that imputing for missing data with the mean value for each patient cohort and postoperative pain severity rating did not change the magnitude of the observed pain differences.

Postoperative Opioid Use

Postoperative opioid requirements were aggregated in two ways: the total postoperative opioid dose was calculated from the sum of all opioids, including buprenorphine, and the supplemental opioid dose requirements were calculated by aggregating full opioid agonists only, excluding buprenorphine. We found patients who continued buprenorphine through surgery had the highest average daily morphine equivalents opioid dose during the first two postoperative weeks (Figure 3A), followed by the group that interrupted buprenorphine and then the control patients (539 mg/day ± 8.5 versus 278 ± 9.8 and 37 ± 0.7 for buprenorphine interrupted and control respectively; P < .001 for pairwise comparisons with Bonferroni correction applied). These results predominately mirror the conversion of buprenorphine to conventional opioids, given the substantial conversion ratio of buprenorphine to morphine (30:1), and the high doses of buprenorphine used to treat opioid use disorder.

Figure 3:

Average and standard deviation of postoperative opioid dose (mg morphine equivalency) in the 2 weeks following surgery (P < 0.001 for pairwise comparisons after Bonferroni correction). A) The mean of the average daily opioid dose for all opioids was highest in the buprenorphine continued cohort, followed by the buprenorphine interrupted group. B) The exclusion of buprenorphine from the analysis shows that patients who continued buprenorphine did not require more supplemental postoperative opioid prescriptions of pure opioid agonists as compared to control (P = 0.33).

Next, we tested the hypothesis that patients who remained on buprenorphine would need significantly higher supplemental dosing with opioid agonists to treat acute pain due to high receptor binding and partial antagonism caused by buprenorphine. We excluded buprenorphine from the aggregate postoperative opioid calculations to examine the amount of supplemental opioid prescription required after surgery (Figure 3B) and found that patients who continued buprenorphine during the perioperative period did not require statistically significantly more supplemental dosing with opioid agonists than the control patients (39.7 mg/day ± 1.9 versus 36.5 ± 0.7 respectively, P = 0.23) and received less supplemental opioids than the group that interrupted buprenorphine (39.7 mg/day ± 1.9 versus 74.2 ± 4.5 respectively; P < 0.001).

Association of Buprenorphine Dose and Other Variables on Postoperative Opioid Prescription

Postoperative supplemental opioid doses (excluding buprenorphine) are plotted as a function of buprenorphine dose for the continued and interrupted cohorts in Figure 4A. Buprenorphine doses above 16 mg/day were associated with higher postoperative opioid prescriptions in the group that interrupted the medication (P-value = .001) but not group that continued it (P-value = .41).

Figure 4:

A) Mean (standard deviation) postoperative opioid dose for patients for patients on different doses of buprenorphine. There was a noticeable dose-response association with buprenorphine over 16 mg/day in the interrupted group. B) The predicted impact of buprenorphine dose on supplemental postoperative opioid doses from a Generalized Additive Model, controlling for postoperative Pain Severity Scale, sex, race, surgical year, and facility complexity rating. There was little impact of buprenorphine doses less than 10 mg/day, which was more pronounced at doses above 16 mg/day.

We performed a sensitivity analysis controlling for various independent variables using a generalized additive model (described below). The preoperative buprenorphine dose was treated as a continuous variable and modeled with a spline curve, showing that postoperative opioid predictions were minimally impacted by buprenorphine doses below 16 mg/day and gradually rose beyond that dose (Figure 4B). The model controlled for other potentially confounding factors, including postoperative day, postoperative pain severity for different procedures, year of surgery, complexity rating of the VA facility, ASA PS Classification, smoking, sex, and race (Supplemental Figure 1).

The study included surgery performed between 2010 and 2020. Procedures performed earlier in the study period (2011 and 2012) were associated with higher predicted opioid requirements (estimate and 95% confidence interval for 2011 was 18.0 [15.3, 20.8], P < 0.001), and surgery dates from 2016 through 2020 were associated with progressively lower requirements (2020 estimate was −12.7 [−15.3, −10.1] mg morphine equivalency, P < 0.001) (Supplemental Digital Content Figure 1). The complexity rating of the VA facility was not associated with a difference in postoperative opioid requirements.

Discussion

Determining the optimal strategy for perioperative buprenorphine management and acute postsurgical pain treatment will become increasingly important as the opioid epidemic evolves and buprenorphine treatment for opioid use disorder becomes more commonplace. There are concerns over the high binding affinity and partial receptor antagonism caused by buprenorphine,²⁸ but recent opinions have shifted to recommending the continuation of buprenorphine in most cases.^20,21 Our study did not find an association between continued buprenorphine administration and higher postoperative pain scores or opioid dose requirements, demonstrating that average pain scores were in a narrow range for all three patient groups, with small differences that were not clinically relevant. Postoperative supplemental opioid utilization was comparable between the group that continued buprenorphine and the opioid-naïve control patients, suggesting that buprenorphine may provide a significant analgesic benefit.

There is concern that buprenorphine may impair the ability to treat acute postoperative pain with a full mu-opioid receptor agonist because of its high mu-opioid receptor affinity and occupancy.^29,30 Receptor binding studies have shown that there was 9 to 20% receptor availability at 16 mg/day, which may allow for sufficient pain control with full mu-opioid agonists when needed without worsened outcomes.^31–33 Our analysis supports the notion that patients can be maintained on buprenorphine after surgery without significantly higher postoperative opioid dose requirements and clinically similar postoperative pain scores.

We examined the time-weighted average postoperative pain scores for 72 hours after surgery and found that they were clinically similar between the three groups. Pain scores were aggregated from inpatient and outpatient clinical encounters, but pain scores were not collected from the Post Anesthesia Recovery and Intensive Care Units due to different methods for documenting vital signs. The time-weighted pain scores were taken from a minimum of 3 postoperative pain scores, and many outpatient surgeries were missing from the analysis. Our results suggest that average pain scores are similar over the first three postoperative days, but we do not have granular data to determine the impact of immediate postoperative pain experienced after emergence from anesthesia.

We included two measurements of postoperative opioid requirements: total postoperative opioids that included buprenorphine in the morphine equivalency calculation and supplemental opioid requirements that excluded it. Total postoperative opioid dosing requirements were highest in the group that continued buprenorphine when buprenorphine was included in the summary statistics; however, this aggregate opioid dose likely reflects the impact of opioid rotation from buprenorphine, including incomplete cross-tolerance and an inexact morphine equivalency for buprenorphine. This aggregate measurement would vary significantly depending on whether a conversion of 10:1 or 30:1 was used for buprenorphine’s morphine equivalent potency. The updated Centers for Disease Control and Prevention opioid guidelines released in 2022 removed buprenorphine from the morphine equivalency table and recommended that buprenorphine not be included in the tabulation of milligrams of morphine equivalency for risk stratification purposes.³⁴ The concept of morphine equivalency can be used to assess relative potency or risk, and while buprenorphine has significant opioid potency, the risk profile is significantly less.

We analyzed the supplemental opioid requirements by excluding buprenorphine from the aggregate measurement to assess opioid utilization and to test the hypothesis that continuation of buprenorphine would not be associated with higher supplemental postoperative opioid agonist requirements in the setting of acute pain. We found that continuing perioperative buprenorphine is not associated with higher postoperative supplemental opioid prescription as compared to control patients without opioid use disorder. When we considered preoperative opioid use in the control group, the group that continued buprenorphine was associated with similar supplemental postoperative opioid requirements to the opioid-naïve control group and significantly lower than the control patients taking an average of ≥ 40 mg of morphine equivalents in the 60 days preceding surgery. This finding is consistent with current knowledge on opioid tolerance and its effect on perioperative opioid requirements.³⁵ These results suggest that buprenorphine may have significant analgesic benefits in acute pain, which may be explained by its relatively high potency, binding affinity, and action at many opioid receptor subtypes.^31,33 The interruption of buprenorphine was associated with higher supplemental opioid prescriptions, which is expected given that patients in that group were taken off buprenorphine and given replacement with pure opioid agonists.

This study’s limitations are typical of large-scale observational studies, including data quality, missing data, and using diagnosis codes as disease markers. We included instances of a patient undergoing more than one procedure, provided the procedures were separated by at least 30 days. While repeat observations may be correlated, the average time between observations was greater than 600 days, which mitigates part of the correlation from repeat observations. Analyzed pain scores were aggregated in the corporate data warehouse from various inpatient and outpatient clinical encounters, but they did not capture pain recorded immediately after surgery due to missing data from the post-anesthesia and intensive care units. This resulted in missing data for patients undergoing outpatient surgery and did not address the severity of pain experienced immediately after surgery. Pain scores are also subjective, and collecting and recording pain score data during routine clinical practice can miss the dynamic nature of acute, postsurgical pain. We examined the impact of missing data by imputing the missing pain scores using the average pain from the cohort and postoperative Pain Severity Scale class and found no change in the magnitude, direction, or significance of the postoperative pain outcome differences. People with opioid use disorder can face bias and stigmatization, which could influence their interactions with the healthcare system, resulting in errors in the documentation of pain scores and undertreatment with opioid agonists in the setting of acute pain. While we cannot validate the recorded, subjective pain intensity scores, there does not appear to be systematic undertreatment of acute pain in the buprenorphine cohorts, given the small differences in average postoperative pain values.

Opioid utilization was assessed by analyzing both inpatient and outpatient pharmacy tables, but certain aspects of perioperative opioid administration were missing, including intraoperative, epidural, and patient-controlled analgesia usage. The missing data represent a significant portion of analgesia provided during the administration of surgical anesthesia. Analysis of outpatient prescription records was limited to the dispensed dosage and intended maximum daily dose, but this may not reflect the intended or actual amount of medication taken by patients or non-medical opioid use. Our study also did not account for the use of regional anesthesia or non-opioid medications used as adjunctive treatment after surgery.

We accounted for variation in the extent of surgical tissue injury and resultant postoperative pain using the postoperative Pain Severity Scale, previously developed from VA data. The postoperative Pain Severity Scale broadly classifies surgical procedure codes into one of 5 classes and may not include important aspects of postoperative pain in all cases. We also found that the year of surgery contributed to predicted postoperative opioid prescription, with predicted supplemental opioid prescriptions decreasing in 2011 and 2012, likely a result of the VA Opioid Safety Initiative, and again declining in 2016, which corresponds with the first publication of the CDC opioid prescribing guidelines.³⁶ We found no significant association between the VA facility complexity rating and predicted postoperative opioid prescription.

We utilized a surrogate measure for determining perioperative buprenorphine management. This method lacks the accuracy of manual chart review but did capture the trend of increasing likelihood of continuing perioperative buprenorphine over the study duration. The method likely misses short-term interruption of buprenorphine and appears to underestimate the rates of buprenorphine interruption reported by Wyse and colleagues.³⁷ Future efforts will focus on using natural language analysis of clinical notes to extract information regarding medication administration instructions to capture buprenorphine management instructions more accurately.

This ad hoc observational study resulted from a preliminary analysis of a VA surgical outcome database, and, as such, it can determine associations but not causal links between perioperative buprenorphine management and postsurgical pain outcomes. The demographics of this population are typical for the Veterans Health Administration but do not generalize to the American population. Future efforts will apply the same methodology to other patient populations to better understand how these results generalize to different populations. Future studies will also focus on surgeries that require postoperative hospital admission so that buprenorphine management can be directly assessed by reviewing the inpatient medication administration record.

Conclusions

Recent guidelines recommend continuing perioperative buprenorphine whenever possible, but clinical outcome evidence is limited. This study suggests that continuing buprenorphine is not associated with higher average pain scores or postoperative opioid agonist requirements, supporting current guideline recommendations. Higher doses of buprenorphine (greater than 16 mg/day) may be associated with rising postoperative opioid requirements.