Effectiveness of Standard Local Anesthetic Bupivacaine and Liposomal Bupivacaine for Postoperative Pain Control in Patients Undergoing Truncal Incisions

Key Points Question Does liposomal bupivacaine reduce postoperative pain and supplemental opioid use more than standard bupivacaine in cardiothoracic and vascular surgical patients? Findings In this randomized clinical trial including 280 participants, no significant difference in pain control was observed between liposomal vs standard formulation bupivacaine. Meaning These results do not support superior performance of liposomal bupivacaine compared with bupivacaine for postoperative pain control in cardiothoracic and vascular truncal incisions.


Introduction
More than 80% of patients undergoing surgical procedures report acute postoperative pain, with less than half achieving adequate postoperative pain control, and nearly 75% of those reporting the severity as moderate, severe, or extreme. [1][2][3][4] This is especially true in open heart, aortic, and lung surgical procedures, where painful truncal incisions are required. Adequate postoperative pain management improves the functional recovery and healing period but also contributes to reduction in postsurgical complication risk and faster patient mobilization, thereby reducing the hospital length of stay and health care costs. 5,6 Short duration of action is a common drawback of most perioperative pain management regimens, including local anesthetic infiltrations lasting for less than 8 hours. 6,7 An injectable extended-release bupivacaine formulation lasting up to 72 hours has gained popularity. Several studies [8][9][10][11][12] on various surgical procedures, including hemorrhoidectomy, bunionectomy, mastectomy, and orthopedic surgery, reported a reduction in postoperative pain (up to 30%) and opioid use following intraoperative use of liposomal bupivacaine compared with placebo and active control. One study 13 integrated the data from 10 randomized, double-blind studies using liposomal bupivacaine via local wound infiltration to assess the efficacy in postoperative pain control and demonstrated substantially prolonged reduction of postsurgical pain, with a greater proportion of patients avoiding use of opioid rescue medication and a lower total opioid consumption over 72 hours in 5 surgical models. A more recent trial 14 showed no difference in opioid use within 48 hours after laparotomy for gynecologic surgery.
Few studies analyze liposomal bupivacaine efficacy in postoperative pain management for major truncal procedures, including vascular, cardiac, laparotomy, and/or thoracic surgical wounds.
One trial 15 evaluated parasternal nerve blockade and found minimal differences between liposomal bupivacaine vs saline. Most studies using long-acting local anesthesia were done for smaller incisions that did not penetrate the chest or abdominal cavities. We conducted a masked, randomized clinical trial to evaluate the effectiveness of liposomal bupivacaine for postoperative pain control following truncal incisions.

Study Design
The study was designed as a randomized, masked, active-controlled, parallel-group clinical trial performed at a single institution between November 2012 and June 2018. The study was approved by the University of Texas Health Science Center at Houston Committee for the Protection of Human Subjects. The study conduct and safety was monitored by an independent data safety monitoring board, composed of 2 surgeons with clinical research master's degrees, an anesthesiologist, and a chaplain, that met periodically during the course of the trial. This study followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline (Figure 1). 16 The trial protocol is shown in Supplement 1.
Because both drug formulations are labeled for use in surgical wound pain control and are in common use for this indication, this was considered to be a comparative effectiveness study, and no support from industry was sought or obtained. Treatment allocation was masked to the patient, the postoperative nursing staff, and the research coordinator conducting the pain and quality of life assessments. Because the appearance of the study drug is different between the liposomal and standard formulations, we did not attempt to mask the treatment group to the surgeon administering the treatment. Most often this was a fellow who had been specifically trained to infiltrate the treatment in a standardized fashion, rather than the attending surgeon. Two separate study coordinators were involved in each case: an unmasked coordinator who obtained the randomization code, consulted with the treating physician, and arranged for the order from the pharmacy, and a masked coordinator who saw the patient daily after surgery and made the postoperative pain scale assessments. Supplemental opioid use was abstracted from the electronic medical record by research personnel masked to group assignment and included all supplemental analgesics delivered by patient-controlled analgesia pump, parenteral injection, or oral route of administration.

Eligibility and Enrollment
Patients aged 18 years or older who required surgery involving 1 of 4 eligible incisions (median sternotomy, laparotomy, thoracotomy, or minithoracotomy) were eligible to participate. Patients were excluded if they had known allergy to bupivacaine or any opioid, or had long-term opioid exposure or a chronic pain disorder that would make them difficult to evaluate for effectiveness of pain control. Conditions that conferred high probability of postoperative morbidity that could interfere with communication of pain status, such as expectation of intubation for more than 24 hours or altered mental status, were also exclusionary. Signed triplicate consent documents were obtained preoperatively, and adequate time was given to allow for patient and family deliberation.
Original documents were included in the physical paper record during the admission. Active participation (assessment of pain and opioid use) was continued for 3 postoperative days.
Complication occurrence was monitored for the entire period of hospitalization.

Administration of Study Drug
Surgery was performed according to routine practice in our group, and no alterations other than treatment with the study drug were made. The 2 treatments were the standard form of bupivacaine hydrochloride (HCl) suspension and a liposomal bupivacaine suspension. Patients in both groups received the same injected volume, 80 mL, divided into 4 20-mL syringes using 22-gauge needles.
The liposomal bupivacaine group received a total dose of 266 mg prepared as one 1.3% 20-mL vial of liposomal bupivacaine diluted in 60 mL of preservative-free normal (0.9%) sterile saline for a total volume of 80 mL. The nonliposomal bupivacaine group received a total dose of 125 mg of bupivacaine HCl prepared as one 0.25% 50 mL or five 0.25% 10-mL vials, diluted in 30 mL of preservative-free normal (0.9%) sterile saline for a total volume of 80 mL. At the time of wound closure, the assigned treatment was infiltrated by injection into the tissue surrounding the wound.

Statistical Analysis
Sample size determination was based on a Cohen effect size of 0.35, which is considered the lower end of the medium effect size range, and for the primary end point would translate to a betweentreatment difference of roughly 2.5 scale points for the area under the curve (AUC). 17 A previous randomized, placebo-controlled trial 12 for pain management following hemorrhoidectomy demonstrated a Cohen effect size of 0.54, considered a large-medium effect, so we used a smaller hypothesized effect size for our active-controlled trial to ensure adequate power. 17 We planned 2 interim analyses using the α spending function of O'Brien and Fleming 18 and determined that a total sample size of 280 evaluable participants would require a final α of 0.0462 to return β = 0.17. The study was not powered to identify differences between incision types, although the randomization was stratified by incision type to ensure balanced treatment allocation within each incision. The randomization schedule was developed using a computer program in blocks of 4 to 6.
Stopping rules were prespecified, with P < .0002 required at the first interim analysis and P < .012 at the second to stop for efficacy. Sample size was calculated using PASS statistical software version 13 (NCSS, LLC). Stopping for safety, had it been necessary, would have been a determination made by the data safety monitoring board.
The primary end point was incisional pain over the first 3 postoperative days as recorded on the Numeric Rating Scale (NRS), an 11-point ordinal scale ranging from 0 (no pain) to 10 (worst pain imaginable). [19][20][21][22] We considered an NRS score greater than 4 as poorly controlled pain and a change in 2 points to be clinically meaningful. 23 Participants were asked to self-aggregate their pain during the previous 24 hours using the NRS. Secondary end points were scores on the Brief Pain Inventory Comparisons were made each day, and the AUC for the NRS over the cumulative 3-day period was also computed using the trapezoidal rule. If pain assessments were missed or patients were discharged before postoperative day 3, imputation of the nonmissing value nearest in time was used (last carried forward method). All supplementary analgesics, including both intravenous and oral opioids, were converted to standard morphine equivalent units (MEU) using a software tool developed in Oregon under a CDC cooperative agreement. 39 Our service generally did not use nonsteroidal anti-inflammatory medications postoperatively because of the prevalence of kidney insufficiency in our patient population. We encouraged Dilaudid as our preferred opioid for breakthrough pain to simplify the analysis, but other opioids were not withheld if prescribed. Daily measures were compared using Wilcoxon rank-sum test and were further assessed for treatmentby-day interaction using nonparametric longitudinal mixed models with unstructured error terms.

JAMA Network Open | Surgery
Main effects of day, treatment, and treatment-by-day interaction were modeled using fixed effects, with a random subject effect to account for within-subject clustering. For these models, P values are computed on ranked dependent variable data, and estimates are modeled using untransformed continuous values. The association between 72-hour pain score and opioid use was analyzed by fixed-effects generalized linear model with interaction. Lengths of stay for intensive care unit and total hospitalization were log-transformed for regression-based analysis but were analyzed by Wilcoxon rank-sum test for univariate comparisons, as were analgesics. If patients could not be assessed for pain because of prolonged intubation and sedation, they were excluded from the length-of-stay analysis. Complication frequencies were compared using contingency table tests, including the χ 2 test where expected value assumptions were met and Fisher exact tests where expected cell frequencies were less than 5. P < .05 was considered significant and all tests were 2-sided. Data were analyzed using SAS statistical software version 9.4 (SAS Institute) from July to December 2018.

Results
We  Table. All patients received their allocated treatment, and there were no follow-up losses in this hospital-based study. The most common reason for exclusion after randomization was prolonged intubation or reintubation after surgery; these patients were sedated such that pain scores could not  Table. In general, according to unpaired daily comparisons, pain scale scores did not differ between groups over the 3-day time period. There was also no difference in satisfaction with pain control on the basis of the 5-point satisfaction questionnaire (Table).       treatment effects of supplemental opioid use did not differ between groups overall, that it did decline significantly over the 3 postoperative day observation period, and that the rate of decline between the treatment groups did not differ (Figure 3). One patient in each treatment group was discharged on postoperative day 2, but both were not taking any opioid pain medications at the time of discharge.
Pain score was associated with supplemental opioid use at all time points and accounted for slightly more than 10% of the variance overall. In general, linear model regression analysis, model terms for effect of pain score (SE) were significant (β = 2.56 [0.55] MEU/NRS unit; P < .001), but treatment group (β = 6.02 [11.12] MEU increase in standard vs liposomal; P = .59) and treatment-bypain interaction (P = .08) were not statistically significant. This indicates that, although opioid use depended on perceived pain, the formulation of bupivacaine administered did not modify this association. In other words, liposomal bupivacaine did not significantly reduce opioid use for a given level of pain compared with standard bupivacaine (Figure 4).
Incision length did not differ between groups, and no incision length-by-treatment interaction was observed. Total NRS-reported pain was higher for thoracotomy incisions (thoracotomy and minithoracotomy combined) than the other incision types (14 [12][13][14][15][16][17]   Standard bupivacaine Liposomal bupivacaine NRS scores are shown by group over 3 days. No main effect of treatment (P = .23) was observed, but significant main effect of day (P < .001) and significant treatment-by-day interaction (P = .03) were present, indicating that rate of pain reduction was greater in the standard bupivacaine group over three postoperative days. Models shown are pain scores; P values are from mixed models of ranked data with unstructured error terms. Lines denotes regression function and shaded areas denote 95% CIs.

Standard bupivacaine Liposomal bupivacaine
Opioid dose (parenteral morphine equivalents) is shown by group over 3 days postoperatively. No main effect of treatment (P = .12) or treatment-by-day interaction (P = .29) was observed, but a significant effect of day (P < .001) was. Hence, reduction in supplemental opioid use over 3 days is significant but does not depend on formulation of bupivacaine. Models shown are opioid doses; P values are from mixed models of ranked data with unstructured error terms. A pairwise contrast at day 1 is statistically significant (P = .04, Wilcoxon rank-sum). Lines denotes regression function and shaded areas denote 95% CIs.
No differences were observed in postoperative complications between the groups (Table).
Neither intensive care unit length of stay nor hospital length of stay was significantly different between groups. There was 1 hospital death, which occurred in the standard formulation group.

Discussion
Effective surgical pain control is an important treatment goal, reduces morbidity, and improves return to activity and to work. [26][27][28] It is also a major patient-centered outcome and an important factor in patient satisfaction and quality of life. Increasingly, development and implementation of opioid-reducing pain management strategies is a substantial public health issue given the scope and scale of the opioid abuse crisis in the US. This is of particular concern for major truncal procedures.
Several recent studies demonstrated that many patients are still using opioids many months after surgery. 29,30 Studies have also shown that postsurgical exposure may increase addiction risk and that even family members of long-term opioid users may be at increased risk for long-term use after their own surgical procedures. [29][30][31][32][33] Improved methods for controlling pain that can minimize opioid use in the postoperative setting are needed, and multimodal nonopioid pain control, including local analgesia, is an important element in a comprehensive pain management strategy. 27,34 Epidural anesthesia can also play a role in certain truncal incisions, but it is not useful for sternotomies or superiorly placed thoracotomy incisions. Epidural anesthesia adversely affects neurological examination after open aortic surgery. Nevertheless, although our service does not routinely use epidural anesthesia, it can be a useful pain control adjunct in selected cases. Placement of thoracic epidural catheters typically does not reside with the surgical team. The advantage of

Standard bupivacaine Liposomal bupivacaine
Opioid consumption is positively correlated with pain (P < .001), with pain accounting for approximately 10% of the variance in opioid use (multiple R 2 = 0.109).
Main effect of treatment is not significant. No modification of the effect by liposomal bupivacaine relative to standard bupivacaine is evident (P for interaction P = .08). Lines denotes regression function and shaded areas denote 95% CIs.
surgeon-administered local anesthesia is that it is fast, easy, and available. We use local anesthesia as part of a successful multimodal regimen that includes nonopioid oral pain medications, gabapentin, locoregional nerve blocks, and dexmedetomidine infusion. 34 Our goal was to determine whether liposomal bupivacaine would improve the intensity and duration of postoperative pain in major truncal surgery as it has been reported to do in other nontruncal orthopedic, cosmetic, and colorectal indications, [8][9][10][11][12][13]15,35 and whether it could also reduce reliance on opioid medications. In this randomized clinical trial involving 280 patients with 4 different types of chest and abdominal incisions, which, to our knowledge, is the largest study of its kind yet to be reported, we were unable to identify any clinically important difference in pain, supplemental opioid use, morbidity, or length of stay between liposomal and standard formulations of bupivacaine.
We did observe significant reductions in pain and opioid use in both groups over 3 postoperative days, and also found that the NRS scores were reduced at a more rapid rate in the bupivacaine HCl group (Figure 2). For major truncal surgery in the setting of a large academic medical center, the findings of this study do not support the hypothesized superiority of liposomal bupivacaine over standard bupivacaine HCl.
The literature on the efficacy of liposomal bupivacaine vs conventionally formulated bupivacaine is equivocal, with multiple publications concluding that liposomal bupivacaine is superior to standard bupivacaine, and others that it is no better. In one case, liposomal bupivacaine was no better even than placebo for the sternotomy indication with respect to supplemental opioid sparing. 15 Reviews in the orthopedic surgery literature 36,37 also concluded that liposomal bupivacaine performed no better than controls. A recent Cochrane review 38

Limitations
There were several limitations in the study. First, in the study design, we anticipated approximately equivalent numbers of incision types. In practice, we performed more sternotomies and fewer thoracotomies and laparotomies. This reflected a global trend toward increasing endovascular repair for thoracic and thoracoabdominal aortic aneurysms with fewer open aortic surgical procedures.
Second, there is variability in technique, speed, and surgeon skill that may make our single-center results not fully replicable in other centers. The absolute pain scores are likely not generalizable because of different pain management regimens used by different groups. However, the comparison and effect size differences will still be useful. Third, although we attempted to standardize the administration of local anesthesia, there may be small differences between surgeons who performed more sternotomies compared with those who performed thoracotomies and laparotomies.
Randomization and stratification by incision type should mitigate these limitations. Fourth, the postoperative pain assessments were done by coordinators masked to treatment assignment, but we could not blind the surgical team administering the local anesthesia. Liposomal bupivacaine has a white, milky appearance whereas standard bupivacaine is clear. Our institutional review board would not allow blinding of the operating room team because of concerns about confusing the study drugs and other drugs with the same appearance, such as propofol. Furthermore, local anesthesia was given at the conclusion of the operation. Some investigators preferred to administer local anesthesia prior to incision. We decided that this would complicate cases where the incision required

JAMA Network Open | Surgery
Standard vs Liposomal Bupivacaine for Postoperative Pain Control in Patients Undergoing Truncal Incisions lengthening for greater exposure. This should not affect the comparison because both groups received anesthesia in the same manner. This study did take longer than anticipated to conduct given the 2-year enrollment hiatus previously described. However, randomization was balanced by blocking every 4 to 6 participants, so any secular trends in pain management would have been absorbed equally into the treatment groups by design.
Fifth, some patients could not be evaluated after randomization because of unanticipated events such as prolonged intubation, so this is not strictly speaking an intent-to-treat analysis. The only randomized patients excluded from the analysis were those who did not have evaluable data, for example, because of prolonged intubation and inability to elicit pain scores. We would have included data for these patients if they existed. As a practical matter, a per protocol analysis is less conservative and, hence, reduces the likelihood of making a type 2 error, which in a negative study such as this would be the greater concern than the type 1 error intent-to-treat protocols are meant to guard against.

Conclusions
The