Efficacy and safety of co-administered bupivacaine and liposomal bupivacaine in interscalene brachial plexus block for shoulder surgery: a systematic review and meta-analysis Authors: Yiming Han, Xingbo Zhang, Weixuan Yao, Songcui Xu, Chan Li, Jun Du Published: 2025 ABSTRACT AbstractBackgroundThe rising incidence of shoulder surgeries highlights significant postoperative challenges, including moderate-to-severe pain, restricted mobility, and opioid-related adverse effects. While interscalene brachial plexus block (ISBPB) utilizing bupivacaine (B) provides effective analgesia, the clinical utility is constrained by a limited duration of action. Liposomal bupivacaine (LB), an extended-release formulation, offers a promising alternative for prolonging analgesia. However, current evidence regarding the comparative efficacy and safety profile remains conflicting. Consequently, a systematic evaluation of the co-administration of B and LB (B + LB) in ISBPB is essential to determine the role of LB in optimizing postoperative pain management for shoulder surgery patients.MethodsThis study included seven randomized controlled trials (RCTs) identified through a comprehensive search of PubMed, Embase, Web of Science, Cochrane Library, and ProQuest databases. Primary outcomes were opioid consumption and multidimensional pain scores (average, resting, activity, and worst pain scores) on each postoperative day (POD). Secondary outcomes included patient satisfaction and adverse events. The meta-analysis was performed using RevMan 5.4, with continuous outcomes analyzed as mean differences (MDs) or standardized mean differences (SMDs), and dichotomous outcomes as risk ratios (RRs). All effect estimates are reported with their 95% confidence intervals (CIs).ResultsCompared to B alone or B with other adjuvant group, the B + LB group exhibited significantly lower opioid consumption on POD 3, and reduced average, resting, and activity pain scores on POD 2. Additionally, the B + LB group showed improvements in worst pain scores on PODs 1–4. Patient satisfaction and adverse event profiles were generally similar between the groups. While an increased risk of hoarseness was specifically observed in the B + LB group on POD 1, the incidence of respiratory complications was similar between the two groups.ConclusionB + LB during ISBPB for shoulder surgery provided superior pain relief on POD 2 and significantly reduced opioid consumption by POD 3, while being associated with an increase in hoarseness risk on POD 1 but comparable overall safety and patient satisfaction.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12871-025-03494-0.Keywords: Shoulder surgery, Interscalene brachial plexus block, Bupivacaine, Liposomal bupivacaine, Meta-analysis Introduction Patients frequently experience moderate-to-severe pain following shoulder surgeries (e.g., total shoulder arthroplasty, rotator cuff repair, and acromioplasty) [1], which may increase the risk of postoperative complications. Brachial plexus nerve blockade is considered optimal analgesia for shoulder surgeries and provides effective pain control [2]. Although bupivacaine (B) is widely used as a long-acting local anesthetic for nerve blocks, the analgesic efficacy of B is inherently limited by a pharmacologic ceiling effect and rarely persists >24 h [3]. Although continuous interscalene brachial plexus block (ISBPB) extends analgesia and reduces opioid requirements compared to single-injection techniques in shoulder surgery [4], the utility of ISBPB is constrained by technical complexity, higher costs, and catheter-related complication risks [5, 6]. In addition, the addition of adjuvants (e.g., epinephrine [E], dexamethasone [D], dexmedetomidine, buprenorphine, and clonidine) to local anesthetics in single injection nerve blocks may variably prolong blockade duration [7]. Reliably and consistently extending ISBPB analgesia beyond 24 h remains challenging [8]. Liposomal bupivacaine (LB; Exparel®; Pacira Pharmaceuticals, Parsippany, NJ, USA) is an extended-release formulation of B encapsulated in multivesicular liposomes that is designed to provide sustained analgesia for up to 72 h following a single administration [9]. The U.S. Food and Drug Administration initially approved LB for surgical site infiltration in 2011 and later expanded the indication for LB to include ISBPB in 2018 [10]. However, the superiority of the co-administration of B and LB (B + LB) over B alone or B with other adjuvants in ISBPB for reducing postoperative opioid consumption and various types of pain scores after shoulder surgery remains controversial [11–17]. This systematic review and meta-analysis comprehensively evaluated the comparative efficacy and safety of B + LB versus B alone or B with other adjuvants for postoperative analgesia following shoulder surgery. Methods This systematic review and meta-analysis were conducted following the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines (Table S1). The study protocol was prospectively registered in the International Prospective Register of Systematic Reviews (PROSPERO; registration ID: CRD42024588867) to ensure methodologic transparency, maintain quality standards, and minimize potential bias. It should be noted that some modifications were made to the original protocol, that require clarification. The title of the project was initially registered in PROSPERO as “The efficacy of bupivacaine combined with bupivacaine liposomes in brachial plexus block for shoulder surgery: a systematic review and meta-analysis”. The title was later revised to the current title to avoid potential misinterpretation by readers, because the original phrasing could imply co-formulation rather than co-administration. The control group definition was revised from plain B to include B alone and B plus adjuvants, ensuring better alignment with current clinical practice in which adjuvants are commonly used in nerve blocks. Secondary outcome measures, including patient satisfaction and adverse events, were added. These additions enhance the comprehensive assessment of the intervention. The subgroup analysis was expanded to include adjuvant use, risk of bias assessment, and local infiltration at the surgical site, enabling a comprehensive assessment of potential sources of heterogeneity. Although not originally planned in the protocol, stratified analyses were performed by postoperative days (PODs) 0, 1, 2, 3, 4 and 7 to better characterize temporal patterns in outcomes. Furthermore, the systematic search was concluded 1 month earlier than the projected timeline (31/12/2024). While some databases updates made it impossible to currently determine exact retrieval numbers of any late 2024 studies, a comprehensive manual review confirmed that no studies qualified for inclusion in the analysis. Search strategy A comprehensive literature search was independently conducted by two researchers (Han Y.M. and Zhang X.B.) using Pubmed, Embase, Web of Science, the Cochrane Library, and Proquest databases from inception to December 2024. The search strategy incorporated the following items: “shoulder surgery”; “shoulder replacement”; “shoulder arthroscopy”; “rotator cuff repair”; “acromioplasty”; “shoulder fracture”; “brachial plexus block”; “bupivacaine”; and “liposomal bupivacaine”. No language restrictions were applied. The complete search strategy and results are presented in Table S2. Inclusion and exclusion criteria The inclusion criteria were as follows: (1) patients aged > 18 years undergoing shoulder surgery under ISBPB; (2) the intervention group received B + LB; (3) the control group received B alone or B plus other adjuvants; and (4) the primary outcomes were required to contain postoperative opioid consumption or at least one type of pain score. The exclusion criteria were as follows: (1) patients who received non-ISBPB or continuous ISBPB; (2) only B or LB administered via surgical site infiltration; (3) inability to obtain complete data after contacting the corresponding authors or to derive valid data estimates through appropriate statistical methods and software tools; (4) letters, comments, case reports, and reviews. Literature screening All the retrieved literature records, including titles, abstracts, and keywords, were imported into Endnote X20 software. Initial duplicate removal was performed using the software automated deduplication function, which identifies duplicates based on the first author’s name, publication date, and article title. The deduplicated records were then subjected to manual verification by two independent investigators (Han Y.M. and Zhang X.B.) to ensure accuracy. Subsequently, the two investigators independently screened the remaining records against the inclusion and exclusion criteria. Discrepancies were resolved through consensus discussion with a third reviewer (Li C.) until unanimous agreement was reached. Risk of bias assessment The methodologic quality of the included trials in the study was evaluated using version 2 of the Cochrane risk-of-bias (RoB-II) tool, which was revised in 2019 and contains the following five domains: randomization; deviation from intended interventions; missing outcome data; outcome measurement bias; and selective reporting [18]. Each trial was categorized as having a “low risk of bias”, “some concerns”, or a “high risk of bias” by two evaluators (Han Y.M. and Yao W.X.). Eventually, all results were checked and disputes were resolved through discussion by a third researcher (Li C.). Data extraction An Excel data extraction form was designed to collect specific information, including general study characteristics, patient demographics, and outcome indicators. The general characteristics and patient demographics included the first author’s name, year of publication, country of study, study design, study groups, age, gender, body mass index, American Society of Anesthesiologists (ASA) grade, surgical types, manufacturer sponsorship, anesthesia methods, anesthetics, and medications used in intervention measures. All pain rating scales applied across studies were explicitly documented. The primary outcome indicators included opioid consumption, average pain scores, resting pain scores, activity pain scores, and worst pain scores. The secondary outcome indicators included patient satisfaction and adverse events. In cases in which direct communication with the authors for obtaining primary data was unattainable, data transformation and estimation methodologies were performed to facilitate the analysis. The estimated means and standard deviations for studies reporting data as medians and interquartile ranges were calculated through a conversion website [19]. Numerical values were extracted and digitized using DigXY software for studies presenting data exclusively in a graphical format, from which corresponding means and standard deviations were subsequently calculated. Two researchers (Han Y.M. and Zhang X.B.) performed the data extraction process independently and discrepancies were resolved by a third reviewer (Li C.). Statistical analysis The meta-analysis was performed using Revman 5.4 and R 4.3.1 statistical software. To address clinical heterogeneity arising from variations in statistical methodologies for opioid use assessment and postoperative pain evaluation scales, standardized mean differences (SMDs) with 95% confidence intervals (CIs) were used for pooling continuous outcome measures. The magnitude of the SMD was interpreted as follows: < 0.2 (trivial effect), 0.2–0.5 (small effect), 0.5–0.8 (moderate effect), and >0.8 (large effect). To enhance clinical interpretability, pooled analyses were also performed using mean differences (MDs) with 95% CIs. A mean difference ≥ 1 point on a 0–10 scale was predefined as the threshold for clinical importance, representing the minimal clinically important difference [20]. For dichotomous adverse events, data were analyzed and presented as risk ratios (RRs) with 95% CIs. All analyses were performed using a random-effects model. Exploratory subgroup analyses were performed to examine the influence of shoulder surgery types, adjuvant use, pain rating scales, risk of bias assessment, and local infiltration at the surgical site on observed heterogeneity. Publication bias was evaluated through multiple approaches: funnel plot symmetry analysis was performed when the number of included studies was ≥ 10; and Egger’s test (for continuous variables) and Peter’s test (for dichotomous variables) were used when the number of included studies ranged from 3 to 10. To ensure the reliability and consistency of the results, sensitivity analyses were performed through sequential exclusion of individual studies. Statistical significance was defined as a P value < 0.05. Results Study selection A total of 306 relevant articles were retrieved from 5 databases: PubMed (n = 128); Embase (n = 34); Web of Science (n = 63); Cochrane Library (n = 25); and ProQuest (n = 56). Following automated removal of 86 duplicates, the titles and abstracts of the remaining articles were screened which excluded 170 articles that were unrelated. The full texts of 50 articles were then assessed for eligibility, ultimately resulting in the inclusion of 7 studies (Fig. 1). Study flow diagram of the included studies General characteristics and patient demographics The study characteristics of the included studies are listed in Table 1. Six RCTs conducted in the U.S. and one in Belgium met the inclusion criteria. These included three studies on total shoulder arthroplasty (TSA) [12, 13, 15], three on non-TSA procedures [11, 14, 16], and one study that encompassed both surgical approaches [17]. Of all included studies, only one received pharmaceutical industry sponsorship [17]. Each study group exhibited comparable demographic characteristics, including age, gender, body mass index (BMI), and ASA classification. The anesthesia methods and anaesthetics used in the included studies are presented in Table 2. All surgical procedures were performed under general anesthesia combined with ISBPB with two studies incorporating additional local infiltration [14, 16]. All ISBPB experimental groups received B + LB, while the control groups were administered either B alone [12, 14, 15, 17] or B with adjuvants (D [11, 16] or D + E [13]). The pharmacologic protocol included B (0.25%–0.5%), LB (133 mg), D (4–5 mg), and E (5 µg). Table 3 summarizes both primary outcomes at 6 time points (postoperative day [POD] 0–4 and 7) and secondary outcomes across all included studies. The risk of bias in the included studies was assessed using the ROB-II tool (Figs. 2 and 3). Briefly, a comprehensive RoB-II assessment of the included studies revealed that 3 studies (42.9%) had some concerns [13–15], while the remaining 4 (57.1%) had a low risk of bias [11, 12, 16, 17]. All opioid consumption data were standardized and reported in milligrams morphine equivalent (MME). Heterogeneous scales were used for pain assessment across studies. One study used a 10-point scale adapted from the overall benefit of analgesia score, also referred to as the modified brief pain inventory (MBPI) [13], two utilized the visual analogue scale (VAS) [11, 15], and four adopted the numerical rating scale (NRS) [12, 14, 16, 17]. Table 1General characteristics and patient demographics of the included studiesFirst author’s nameYearCountryDesignGroups (n)*Age (y)*Gender (n, M/F)BMI (kg/m2)*ASA grade*Surgery typesManufacturer sponsorshipBaessler et al. [11]2020AmericaRCTB + D (26)B + LB (24)59.1 ± 9.056.9 ± 9.615/1114/10N/SN/Srotator cuff repairNoneElmer et al. [12]2023AmericaRCTB (40)B + LB (40)67 ± 0.167 ± 0.119/2124/1630.2 ± 0.231.5 ± 0.2II/III/IV 1/22/17II/III/IV 0/20/20total shoulder arthroplastyNoneFinkel et al. [13]2024AmericaRCTB + D + E (45)B + LB (45)73 (66.5, 78.0)72 (66.5, 77.0)22/2323/2231.6 (27.0, 35.9)31.3 (27.3, 33.7)2 (2,3)2 (2,3)total shoulder arthroplastyNoneFlaherty et al. [14]2022AmericaRCTB (35)B + LB (35)57.6 ± 8.859.2 ± 1124/1122/1331.7 ± 7.0230.8 ± 5.06I/II/III 7/14/14I/II/III 6/16/13rotator cuff repairNoneHattrup et al. [15]2021AmericaRCTB (52)B + LB (52)69.2 ± 10.1570.0 ± 6.8428/2430/2229.2 ± 5.4629.4 ± 3.98I/II/III 1/36/15I/II/III 2/39/11total shoulder arthroplastyNoneKim et al. [16]2022AmericaRCTB + D (56)B + LB (55)49 (39, 63)55 (48, 61)48/843/1227.5 ± 4.127.6 ± 4.1I/II 12/44I/II 10/45arthroscopic shoulder surgeryNoneVandepitte et al. [17]2017BelgiumRCTB (24)B + LB (26)57 ± 1261 ± 1110/1413/1327 ± 528 ± 4I/II/III 7/17/0I/II/III 6/19/1total shoulder arthroplasty orrotator cuff repairYesn Number, y Years, M Male, F Female, BMI Body mass index, ASA American Society of Anesthesiologists, RCT Randomized controlled trial, B Bupivacaine, LB Liposomal bupivacaine, D Dexamethasone, E Epinephrine, N/S Not specified*Data are presented as mean ± SD (normally distributed), median [interquartile range, Q1-Q3] (non-normally distributed), and/or counts for categorical variables General characteristics and patient demographics of the included studies B + D (26) B + LB (24) 59.1 ± 9.0 56.9 ± 9.6 15/11 14/10 B (40) B + LB (40) 67 ± 0.1 67 ± 0.1 19/21 24/16 30.2 ± 0.2 31.5 ± 0.2 II/III/IV 1/22/17 II/III/IV 0/20/20 B + D + E (45) B + LB (45) 73 (66.5, 78.0) 72 (66.5, 77.0) 22/23 23/22 31.6 (27.0, 35.9) 31.3 (27.3, 33.7) 2 (2,3) 2 (2,3) B (35) B + LB (35) 57.6 ± 8.8 59.2 ± 11 24/11 22/13 31.7 ± 7.02 30.8 ± 5.06 I/II/III 7/14/14 I/II/III 6/16/13 B (52) B + LB (52) 69.2 ± 10.15 70.0 ± 6.84 28/24 30/22 29.2 ± 5.46 29.4 ± 3.98 I/II/III 1/36/15 I/II/III 2/39/11 B + D (56) B + LB (55) 49 (39, 63) 55 (48, 61) 48/8 43/12 27.5 ± 4.1 27.6 ± 4.1 I/II 12/44 I/II 10/45 B (24) B + LB (26) 57 ± 12 61 ± 11 10/14 13/13 27 ± 5 28 ± 4 I/II/III 7/17/0 I/II/III 6/19/1 total shoulder arthroplasty or rotator cuff repair n Number, y Years, M Male, F Female, BMI Body mass index, ASA American Society of Anesthesiologists, RCT Randomized controlled trial, B Bupivacaine, LB Liposomal bupivacaine, D Dexamethasone, E Epinephrine, N/S Not specified *Data are presented as mean ± SD (normally distributed), median [interquartile range, Q1-Q3] (non-normally distributed), and/or counts for categorical variables Table 2Details of anesthesia methods and drugs of the included studiesFirst author’s nameAnesthesia methodsAnestheticsOndansetronDexamethasoneNSAIDsMidazolamGAISBPBLIBaessler et al. [11]GA + ISBPBsevoflurane or desfluraneB + D: 0.5% bupivacaine + 4 mg dexamethasoneB + LB: 0.5% bupivacaine + 133 mg liposomal bupivacaineNoneYesYesYesN/SElmer et al. [12]GA + ISBPB1–2 mg/kg propofol; 0.6 mg/kg rocuronium; 0.5 mg/kg ketamineB: 0.5% bupivacaineB + LB: 0.5% bupivacaine + 133 mg liposomal bupivacaineN/SN/SYesYesYesFinkel et al. [13]GA + ISBPB1–2 mg/kg propofol; 1–3 µg/kg fentanyl; 0.5–1.5 mg/kg lidocaine; 0.5–1.5 mg/kg rocuroniumB + D + E: 0.5% bupivacaine + 5 mg dexamethasone + 5 µg epinephrineB + LB: 0.5% bupivacaine + 133 mg liposomal bupivacaineN/SN/SYesYesYesFlaherty et al. [14]GA + ISBPB + LI1.5–3.5 mg/kg propofol; 60–100 mg lidocaine; 0.25 mg/kg ketamine; muscle relaxants or fentanylB: 0.5% bupivacaineB + LB: 0.5% bupivacaine + 133 mg liposomal bupivacaine10 ml 0.5% bupivacaineYesYesYesYesHattrup et al. [15]GA + ISBPBpropofol; fentanyl; rocuronium or succinylcholine; sevofluraneB: 0.5% bupivacaineB + LB: 0.25% bupivacaine + 0.5% bupivacaine + 133 mg liposomal bupivacaineN/SYesYesYesYesKim et al. [16]GA + ISBPB + LI2 mg/kg propofol; 20-200ug fentanylB + D: 0.5% bupivacaine + 4 mg dexamethasoneB + LB: 0.5% bupivacaine + 133 mg liposomal bupivacaine5–10 ml local anestheticYesYesYesYesVandepitte et al. [17]GA + ISBPB3–4 µg/kg propofol; 1–2 µg/kg/min remifentanil; 0.6 mg/kg rocuroniumB: 0.25% bupivacaineB + LB: 0.25% bupivacaine + 133 mg liposomal bupivacaineN/SN/SYesYesN/SGA General anesthesia, ISBPB Interscalene brachial plexus block, LI Local infiltration, N/S Not specified, B Bupivacaine, LB Liposomal bupivacaine, D Dexamethasone, E Epinephrine Details of anesthesia methods and drugs of the included studies B + D: 0.5% bupivacaine + 4 mg dexamethasone B + LB: 0.5% bupivacaine + 133 mg liposomal bupivacaine B: 0.5% bupivacaine B + LB: 0.5% bupivacaine + 133 mg liposomal bupivacaine B + D + E: 0.5% bupivacaine + 5 mg dexamethasone + 5 µg epinephrine B + LB: 0.5% bupivacaine + 133 mg liposomal bupivacaine B: 0.5% bupivacaine B + LB: 0.5% bupivacaine + 133 mg liposomal bupivacaine B: 0.5% bupivacaine B + LB: 0.25% bupivacaine + 0.5% bupivacaine + 133 mg liposomal bupivacaine B + D: 0.5% bupivacaine + 4 mg dexamethasone B + LB: 0.5% bupivacaine + 133 mg liposomal bupivacaine B: 0.25% bupivacaine B + LB: 0.25% bupivacaine + 133 mg liposomal bupivacaine GA General anesthesia, ISBPB Interscalene brachial plexus block, LI Local infiltration, N/S Not specified, B Bupivacaine, LB Liposomal bupivacaine, D Dexamethasone, E Epinephrine Table 3Outcome indicators of the included studiesFirst author’s namePrimary outcomesSecondary outcomesOpioid consumption(MME)Resting pain scoresActivity pain scoresWorst pain scoresAverage pain scoresPatientsatisfactionAdverseeventsPOD012347Pain scales012347012347012347012347Baessler et al. [11]••••VAS••••Elmer et al. [12]••••NRS••••••••••••••••Finkel et al. [13]•••MBPI•••••••••••Flaherty et al. [14]•••NRS••••Hattrup et al. [15]VAS•••Kim et al. [16]••••••NRS•••••••••••••••••••Vandepitte et al. [17]•••••NRS•••••••MME Milligram morphine equivalent, POD Postoperative day, VAS Visual analogue scale, NRS Numerical rating scale, MBPI Modified brief pain inventory Outcome indicators of the included studies MME Milligram morphine equivalent, POD Postoperative day, VAS Visual analogue scale, NRS Numerical rating scale, MBPI Modified brief pain inventory Risk of bias assessment for the included studies Summary plot for risk of bias assessment for the included studies Primary outcomes Opioid consumption A comparative analysis of opioid consumption between the B and B + LB groups was performed on POD 0, 1, 2, 3, and 7. The pooled analysis of six studies [11–14, 16, 17] on POD 3 demonstrated a statistically significant reduction in opioid consumption in the B + LB group (SMD = − 0.53, 95% CI = − 0.77 to − 0.28; P < 0.0001, I² = 37%; MD = − 2.87, 95% CI = − 4.83 to − 0.91; P = 0.004, I² = 67%). The sensitivity analysis of opioid consumption on POD 3 revealed the following: exclusion of the study by Elmer et al. [12] in the SMD analysis and the study by Vandepitte et al. study [17] in the MD analysis reduced heterogeneity to 0%, without affecting the overall results. In contrast, meta-analyses at other time points (PODs 0, 1, 2, 4, and 7) showed no significant differences in opioid requirements between the groups. Removal of the Vandepitte et al. study [17], which was sponsored by pharmaceutical companies, significantly affected the outcome on POD 2 (n = 5 [number of included studies], SMD = − 0.59, 95% CI = − 1.00 to − 0.19; P = 0.004, I² =74%; MD = − 5.28, 95% CI = − 6.75 to − 3.82; P < 0.00001, I² =0%). Subgroup analysis revealed that surgical type (MD) as a significant source of heterogeneity in POD3 opioid consumption, whereas adjuvant use, risk of bias assessment, and local infiltration at the surgical site did not significantly affect MD or SMD heterogeneity. Additionally, the Egger’s test indicated the absence of significant publication bias on PODs 1–4 (Table 4 and Files S1-6). Table 4Pooled primary outcomes of different time pointsOutcomesStudiesEffect modelHeterogeneityI2 (%)SMDHeterogeneityI2 (%)MDEgger’s testEffect size (95% CI)P valueEffect size (95% CI)P valueOpioid consumption POD 02Random0−0.10 [−0.38, 0.19]0.500−0.45 [−1.40, 0.50]0.35N/A POD 16Random69−0.21 [−0.55, 0.13]0.2279−1.61 [−3.78, 0.56]0.150.65 POD 26Random89−0.31 [−0.90, 0.29]0.3193−5.08 [−9.64,−0.52]0.03#0.65 POD 36Random37−0.53 [−0.77,−0.28]< 0.000167−2.87 [−4.83,−0.91]0.004*0.30 POD 43Random0−0.18 [−0.46, 0.09]0.180−0.23 [−0.73, 0.27]0.380.89 POD 72Random0−0.03 [−0.34, 0.28]0.830−0.18 [−0.65, 0.28]0.44N/AAverage pain score POD 02Random710.09 [−0.43, 0.61]0.74750.17 [−0.59, 0.93]0.66N/A POD 14Random0−0.02 [−0.23, 0.19]0.840−0.01 [−0.42, 0.39]0.950.44 POD 24Random18−0.53 [−0.77,−0.29]< 0.000143−1.17 [−1.78,−0.56]0.0002*0.80 POD 34Random70−0.32 [−0.71, 0.08]0.1273−0.66 [−1.50, 0.19]0.13#0.16 POD 43Random85−0.20 [−0.88, 0.47]0.5586−0.39 [−1.63, 0.85]0.53#0.34Resting pain score POD 02Random470.03 [−0.36, 0.42]0.88500.04 [−0.53, 0.62]0.88N/A POD 12Random0−0.32 [−0.61,−0.03]0.0317−0.46 [−0.98, 0.06]0.08N/A POD 22Random21−0.78 [−1.12,−0.44]< 0.0000145−1.12 [−1.76,−0.48]0.0006*N/A POD 32Random0−0.35 [−0.64,−0.06]0.0232−0.49 [−1.03, 0.06]0.08N/A POD 72Random0−0.17 [−0.45, 0.12]0.2543−0.21 [−0.77, 0.35]0.47N/AActivity pain score POD 12Random47−0.28 [−0.68, 0.12]0.170−0.61 [−1.15,−0.07]0.03N/A POD 22Random83−0.86 [−1.61,−0.11]0.020−1.14 [−1.51,−0.78]< 0.00001*N/A POD 32Random0−0.53 [−0.82,−0.24]0.000458−0.79 [−1.54,−0.03]0.04N/A POD 72Random870.09 [−0.72, 0.91]0.8283−0.08 [−1.34, 1.19]0.90N/AWorst pain score POD 14Random93−1.06 [−2.02,−0.11]0.0367−1.33 [−1.96,−0.70]< 0.0001*0.04 POD 24Random91−1.34 [−2.24,−0.44]0.00471−1.70 [−2.32,−1.08]< 0.00001*0.12 POD 34Random91−1.26 [−2.14,−0.38]0.00587−1.40 [−2.19,−0.61]0.0005*0.04 POD 42Random77−1.04 [−1.83,−0.24]0.0145−1.11 [−1.78,−0.44]0.001*N/A POD 72Random96−0.59 [−2.62, 1.45]0.5796−0.31 [−1.68, 1.07]0.66N/ASMD Standardized mean difference, MD Mean difference, CI Confidence interval, POD Postoperative day, N/A Not applicable#Removal of one study, an SMD > 0.5, an MD > 1.0, and P value < 0.05*An SMD > 0.5, an MD > 1.0, and P value < 0.05 Pooled primary outcomes of different time points SMD Standardized mean difference, MD Mean difference, CI Confidence interval, POD Postoperative day, N/A Not applicable #Removal of one study, an SMD > 0.5, an MD > 1.0, and P value < 0.05 *An SMD > 0.5, an MD > 1.0, and P value < 0.05 Average pain scores Average pain scores were compared between the B and B + LB groups at five postoperative time points (PODs 0–4). Pooled analysis demonstrated comparable scores between the groups on PODs 0, 1, 3, and 4. However, both effect size measures (n = 4, SMD = − 0.53, 95% CI = − 0.77 to − 0.29; P < 0.0001, I² = 18%; MD = − 1.17, 95% CI = − 1.78 to − 0.56; P = 0.0002, I² = 43%) indicated lower average pain scores in the B + LB group at POD 2. In addition, the Baessler et al. study [11] was the main source of heterogeneity, which resulted in a remarkable reduction in heterogeneity and affected the outcome after exclusion on POD 3 (n = 3, SMD = − 0.51, 95% CI = − 0.74 to − 0.28; P < 0.0001, I² = 0%; MD = − 1.08, 95% CI = − 1.56 to − 0.61; P < 0.00001, I² = 0%) and POD 4 (n = 2, SMD = − 0.56, 95% CI = − 0.85 to − 0.27; P = 0.0001, I² = 0%; MD = − 1.04, 95% CI = − 1.55 to − 0.52; P < 0.0001, I² = 0%). Subgroup analyses revealed that adjuvant use, surgery types, pain rating scale, risk of bias assessment, and local infiltration at the surgical site did not serve as significant sources of heterogeneity for MD or SMD outcomes at POD 2. The Egger’s test showed that publication bias was not significant on PODs 1–4 (Table 4 and Files S1-6). Resting pain scores The meta-analysis showed a statistically significant reduction in resting pain scores for LB addition on POD 2 (n = 2, SMD = − 0.78, 95% CI = − 1.12 to − 0.44; P < 0.00001, I² = 21%; MD = − 1.12, 95% CI = − 1.76 to − 0.48; P = 0.0006, I² = 45%) when compared to the control group. However, no significant differences were reported between the two groups at other time points (PODs 0, 1, 3, and 7). Subgroup analyses revealed that adjuvant use and surgery types did not contribute significantly to heterogeneity in resting pain scores at POD2. However, the pain rating scale, risk of bias assessment, and local infiltration at the surgical site could not be analyzed by subgroup due to insufficient stratification in the included studies (Table 4 and Files S1-6). Activity pain scores The meta-analysis of activity pain scores revealed a statistically significant reduction in the B + LB group on POD 2 (n = 2, SMD = − 0.86, 95% CI = − 1.61 to − 0.11; P = 0.02, I² = 83%; MD = − 1.14, 95% CI = − 1.51 to − 0.78; P< 0.00001, I² = 0%) with no significant differences observed on PODs 1, 3 and 7. Subgroup analysis identified adjuvant use (SMD) as a significant source of heterogeneity in POD2 activity pain scores, whereas surgical type did not significantly affect MD or SMD heterogeneity. Subgroup analysis of pain rating scales, risk of bias assessment, and local infiltration at the surgical site was precluded due to insufficient stratified reporting in the primary studies (Table 4 and Files S1-6). Worst pain scores The reduction in the B + LB group worst pain scores reached statistical significance on PODs 1 (n = 4, SMD = − 1.06, 95% CI = − 2.02 to − 0.11; P = 0.03, I² = 93%; MD = − 1.33, 95% CI = − 1.96 to − 0.70; P < 0.0001, I² = 67%), 2 (n = 4, SMD = − 1.34, 95% CI = − 2.24 to − 0.44; P = 0.004, I² = 91%; MD = − 1.70, 95% CI = − 2.32 to − 1.08; P < 0.00001, I² = 71%), 3 (n = 4, SMD = − 1.26, 95% CI = − 2.14 to − 0.38; P = 0.005, I² = 91%; MD = − 1.40, 95% CI = − 2.19 to − 0.61; P = 0.0005, I² = 87%), and 4 (n = 2, SMD = − 1.04, 95% CI = − 1.83 to − 0.24; P = 0.01, I² = 77%; MD = − 1.11, 95% CI = − 1.78 to − 0.44; P = 0.001, I² = 45%), while no significant between-group differences were detected on POD 7. Exclusion of the Vandepitte et al. study [17], which was sponsored by pharmaceutical companies, resulted in a substantial reduction in heterogeneity for worst pain scores on POD 3 with no impact on outcome. Subgroup analyses identified surgery types as a potential source of heterogeneity for SMD or MD in worst pain scores during PODs 1–4. Additionally, adjuvant use, pain rating scale, risk of bias assessment, and local infiltration at the surgical site may contribute to heterogeneity at some postoperative time points. The Egger’s test revealed potential publication bias in the worst pain score on PODs 1 and 3 (Table 4 and Files S1-6). Secondary outcomes Patient satisfaction Three studies [12, 13, 16] evaluated patient satisfaction on POD 0. While one study [16] reported lower satisfaction scores with LB, the remaining two studies [12, 13] detected no statistically significant differences between the groups. Four studies [12, 13, 16, 17] assessed patient satisfaction on PODs 1, 2, and 3. No significant differences between the groups were observed on PODs 1 and 3 in any study. However, on POD 2 one study [17] reported improved patient satisfaction while the remaining 3 studies [12, 13, 16] detected no differences. Among the three studies [13, 16, 17] assessing POD 4 patient satisfaction, LB administration demonstrated a significant improvement in one study [13], while the remaining two studies [16, 17] showed comparable patient satisfaction scores between the groups. Three studies [12, 16, 17] evaluated patient satisfaction on POD 7 and showed no difference between the groups. Adverse events Five studies [12–14, 16, 17] reported adverse events with generally comparable safety profiles between the groups. The notable difference was an increased incidence of hoarseness in the B + LB group on POD 1 (LB: 25 cases vs. B: 14 cases; P = 0.014) [16] and two cases of dyspnea attributed to the block [14]. However, there was no statistically significant difference in unexpected admissions for shortness of breath following surgery between the groups (LB: 2 cases vs. B: 1 case; P = 0.6) [12]. One study [14] reported a higher overall complication rate with LB (LB: 9 cases vs. B: 2 cases; P = 0.05), 2 other studies [13, 17] detected no significant differences in adverse event rates. Neurologic safety was comparable (LB: 6 cases vs. B: 3 cases; P = 0.483) on POD 7 with no persistent symptoms on POD 14 [16]. A pooled analysis for specific time points was feasible only for POD 1, as just three studies reported on hoarseness and respiratory complications with time-specific data for that day [12, 16, 17]. Pooled analysis indicated a significantly increased risk of hoarseness in the B + LB group compared to the control (n = 2, RR = 1.88, 95% CI = 1.23 to 2.87; P = 0.004, I² = 0%). However, the risk of respiratory complications was comparable between the two groups (n = 3, RR = 0.83, 95% CI = 0.23 to 3.03; P = 0.78, I² = 31%; File S1). Discussion This systematic review and meta-analysis demonstrated that B + LB in ISBPB for shoulder surgery statistically reduced opioid consumption on POD 3 and maximized pain relief across all measured pain scores on POD 2, including average, resting, activity, and worst pain. Furthermore, patient satisfaction and the overall profile of adverse events were comparable between the B + LB group and the groups receiving B alone or B with other adjuvants, apart from an increased risk of hoarseness on POD 1. Postoperative pain following shoulder surgery is typically moderate-to-severe in intensity [21], arising from a complex interplay of surgical factors, anesthetic management, and individual patient characteristics. Rotator cuff repair and shoulder arthroplasty represent the two most prevalent surgical interventions for managing shoulder dysfunction but demonstrate comparable early postoperative pain profiles despite significant differences in procedural invasiveness [22]. This apparent paradox extends to surgical approaches as well. While arthroscopic techniques offer pain reduction benefits through minimized tissue disruption, clinical evidence does not support statistically significant differences in postoperative pain levels when compared to open surgical repair techniques [23, 24]. Notably, both ISBPB and surgical site infiltration have been shown to significantly reduce postoperative pain following shoulder surgery, although the comparative efficacy remains debatable [25]. Given the limitations of continuous ISBPB, adding adjuvants to local anesthetics in ISBPB represents an alternative approach to prolong blockade duration. Among these adjuvants, D has optimal prolongation effects, extending analgesia by 6–8 h [26]. However, compared to continuous ISBPB, these adjuvants rarely achieve a 24-h duration of blockade. The use of B alone, B + D, or B + D + E in the control group unlikely affected the outcome indicators on POD 1. Consistent with this, adjuvant use was generally not a major source of heterogeneity in the primary outcomes. In contrast, surgical type appeared to be the predominant contributor to heterogeneity in the worst pain scores. LB is a sustained-release formulation that encapsulates B in multivesicular liposomes [27], producing biphasic drug release (peaks at 1–2 h and 24–48 h) [28, 29]. Although LB alone has not consistently outperformed short-acting local anesthetics [30–32], B + LB may offer immediate and prolonged analgesia. However, the clinical results are conflicting. Our meta-analysis found that B + LB most effectively reduced pain scores on POD 2 and opioid use on POD 3, likely due to the 24-h half-life of opioids delaying dosage adjustments. While Chevrollier et al. [33] reported better day-of-surgery analgesia with B + LB in transversus abdominis plane (TAP) for colorectal surgery, Hamilton et al. [34] reported that B + LB administration did not confer any measurable benefit in patients undergoing knee replacement surgery with periarticular injection. Similarly, Chan et al. [35] observed early analgesia (POD 1) with B + LB in supraclavicular brachial plexus blockade for forearm surgery, whereas our data showed peak effects later (POD 2). Such variability may stem from technical factors (e.g., injection accuracy [36], tissue diffusion [37], local inflammation [31], or vascular effects on drug release [38]), surgical site differences, or study design. Notably, B + LB reduced opioid use in TAP for cesarean delivery [39] and NSAID use in pudendal nerve block for colporrhaphy [40]. While some studies reported statistically significant recovery differences [14, 41], the clinical relevance remains uncertain. While plasma levels of local anesthetics may not accurately reflect the amount available at the injection site for nerve blockade due to variable local vascularity affecting systemic absorption, the local pharmacologic effect is not necessarily correlated with plasma concentration. Consequently, the duration of analgesia cannot be estimated from the detection time of B in the blood [42]. However, the systemic pharmacokinetic effects assessed via plasma levels should relate to the B local release [17]. As previously noted, the B plasma concentrations peak at 24–48 h in the B + LB group, which was consistent with our observation of reduced postoperative pain across various types on POD 2. In addition, safety concerns also exist regarding the use of LB as an adjuvant to non-liposomal local anesthetics. Mixing LB with other solutions can accelerate B release, potentially increasing the risk of systemic toxicity. Nevertheless, clinical evidence has shown that surgical site infiltration or nerve blocks using B (50/100/125 mg) + LB (133/266 mg) do not produce plasma concentrations reaching the toxic threshold for B (2 µg/mL) [28, 43]. Mixtures are considered safe provided the milligram dose ratio of plain B solution-to-LB does not exceed 1:2 [37]. The present analysis found no significant differences in respiratory or other major complications on POD1, despite a higher incidence of hoarseness with B + LB. This conclusion is limited by the small number of studies assessing these specific outcomes, and therefore requires validation in larger, dedicated investigations. One of the included studies has received funding or support from pharmaceutical companies producing LB [17], which could introduce bias in reporting favorable outcomes [44]. However, excluding this study not only maintained the positive outcomes but actually strengthened the results, yielding statistically significant findings for opioid consumption on POD2. We therefore conclude that industry sponsorship had no substantial impact on our overall conclusions. Several limitations of this study should be discussed. First, Although grey literature inclusion is recommended to minimize publication bias, our search strategy did not encompass dissertations, conference proceedings, or preprint platforms, which may affect the review exhaustiveness. Second, the limited number of studies in some comparisons resulted in low statistical power for tests of publication bias (e.g., Egger’s test), warranting a cautious interpretation of these results. Third, the sample sizes were small across all included studies. Fourth, the inherent physical properties of LB, such as its milky-white appearance and high viscosity, make true blinding challenging in practice. Fifth, differences in the types and dosages of local anesthetics across all groups may also affect the results. Sixth, although this meta-analysis demonstrated that LB offers advantages in reducing opioid consumption and improving analgesic efficacy at specific time points, the available data are insufficient to adequately assess the cost-effectiveness. Finally, the predominance of studies from the U.S. may limit the direct applicability of our conclusions to other healthcare settings, where factors such as standardized ERAS pathways, opioid-prescribing customs, and financial reimbursement structures likely differ. Conclusions B + LB during ISBPB for shoulder surgery significantly reduced opioid consumption on POD 3 and provided superior analgesia across all pain metrics on POD 2. Apart from an increase in the risk of hoarseness on POD 1, the overall incidence of complications and patient satisfaction remained comparable between groups. Supplementary Information Supplementary Material 1: Preferred Reporting Items for Systematic Reviews and Meta-analyses checklist. Supplementary Material 2: Detailed description of search strategies and results. Supplementary Material 3: Forest plot analysis of outcomes. Supplementary Material 4: Subgroup analysis of surgery type. Supplementary Material 5: Subgroup analysis of adjuvant use. Supplementary Material 6: Subgroup analysis of pain rating scale. Supplementary Material 7: Subgroup analysis of risk of bias assessment. Supplementary Material 8: Subgroup analysis of local infiltration at the surgical site. Acknowledgements Not applicable. Abbreviations American Society of Anesthesiologists Bupivacaine Co-administered bupivacaine and liposomal bupivacaine Body mass index Confidence interval Dexamethasone Epinephrine Female General anesthesia Interscalene brachial plexus block Liposomal bupivacaine Local infiltration Male Modified brief pain inventory Milligrams morphine equivalent Number Not applicable Not specified Numerical rating scale Postoperative day Preferred Reporting Items for Systematic Reviews and Meta-analyses Randomized controlled trial Version 2 of the Cochrane risk-of-bias Risk ratios Standardized mean difference Transversus abdominis plane Total shoulder arthroplasty Visual analogue scale Year Authors’ contributions Han Y.M., Zhang X.B., Yao W.X., Xu S.C., Li C. and Du J. conceived and designed the study, collected and analyzed data, and drafted the manuscript. Han Y.M. and Du J. contributed to statistical analysis and manuscript revision. All authors reviewed and approved the final manuscript for submission. Funding Not applicable. Data availability All relevant data are included in the manuscript and supplementary information files. Declarations Ethics approval and consent to participate Not applicable. This study was exempt from ethics approval by the Ethics Committee because the study involved a systematic review of existing literature. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1: Preferred Reporting Items for Systematic Reviews and Meta-analyses checklist. Supplementary Material 2: Detailed description of search strategies and results. Supplementary Material 3: Forest plot analysis of outcomes. Supplementary Material 4: Subgroup analysis of surgery type. Supplementary Material 5: Subgroup analysis of adjuvant use. Supplementary Material 6: Subgroup analysis of pain rating scale. Supplementary Material 7: Subgroup analysis of risk of bias assessment. Supplementary Material 8: Subgroup analysis of local infiltration at the surgical site. Data Availability Statement All relevant data are included in the manuscript and supplementary information files.