In this prospective, double-blind, randomized controlled trial of 61 patients undergoing lumbar spine surgery, supplementing ropivacaine with epinephrine for bilateral ultrasound-guided ESPB produced a statistically significant difference in the primary endpoint, with lower NRS pain scores at PACU admission in the intervention group (median, 4.0 vs 5.0; P = 0.0038; r = 0.42).
The 1-point absolute reduction is below the commonly cited MCID of approximately 1.5 - 2 points for acute pain (
24), and the appropriateness of this threshold for pain measured immediately after surgery in the PACU is itself uncertain. The threshold is derived largely from chronic pain and emergency department settings and may not be directly transferable to acute postoperative pain, although some postoperative studies suggest a threshold closer to 1 point. The observed difference may therefore lie, at best, at the lower boundary of clinical relevance. The fully adjusted analysis was borderline (β = −1.250, 95% CI −2.47 to −0.028; P = 0.045), with a wide confidence interval approaching the null. Accordingly, the trial should be regarded as negative for its primary, clinically anchored endpoint, with the statistically significant difference reported for completeness; a clinically unimportant effect cannot be excluded. The secondary outcomes are exploratory and cannot substitute for a clinically important effect on the primary endpoint. Although sensory block duration was not measured, the lower intraoperative opioid consumption and longer time to first rescue analgesia are consistent with prolonged blockade, but this mechanism is inferred rather than demonstrated.
The intervention group received less intraoperative fentanyl and was estimated to have consumed less isoflurane. These findings may indicate improved intraoperative analgesia, but several factors complicate their interpretation. Surgical duration differed substantially between groups (mean, 232.3 min vs 161.3 min), and longer procedures both increase total anesthetic consumption and provide more opportunities for fentanyl administration. Fentanyl was analyzed as total dose only, and adjustment for duration in a regression model does not remove confounding when the outcome is cumulative. Fentanyl administration was triggered by hemodynamic thresholds; therefore, consumption may reflect factors other than nociceptive input. Isoflurane consumption was estimated from vaporizer settings and duration rather than measured by gas analysis. The estimate assumes constant fresh gas flow and vaporizer output, does not account for temperature and pressure effects or circuit losses, and depends on a conversion method that was not documented in sufficient detail for independent replication. We did not measure emergence times or recovery outcomes, and the lower estimated isoflurane consumption should not be interpreted as evidence of faster emergence or improved recovery. These secondary, exploratory findings were unadjusted for multiplicity and confounded by surgical duration; they are not robust enough on their own to support a clinical benefit beyond the primary pain outcome and require confirmation in larger studies with standardized protocols, direct gas analysis, and normalization to body weight and surgical duration.
A major limitation is the absence of cumulative postoperative opioid consumption data at 12 or 24 hours. Although time to first opioid request was markedly prolonged (median, 18 h vs 0 h; P < 0.0001), this does not necessarily indicate reduced total opioid exposure: intervention patients may have required substantial rescue analgesia once the block resolved, particularly given their higher 12-hour pain scores. Intraoperative fentanyl consumption and postoperative opioid consumption are distinct outcomes with different clinical implications, and only the former was measured. The opioid-sparing claim is therefore restricted to the intraoperative period, and future studies should include cumulative opioid consumption at 12, 24, and 48 hours.
Our results align with the broader literature on adjuvant use in ESPB and provide data specifically on epinephrine augmentation. Ye and colleagues analysed 15 RCTs (1,035 patients) and found that ESPB reduced intraoperative opioid use by a mean difference of −31.70 mg morphine equivalents (95% CI −41.69 to −21.71) and 24-hour postoperative opioid consumption by −11.02 mg (95% CI −14.42 to −7.63), and reduced the risk of postoperative nausea and vomiting (
14); a 2025 umbrella review reached concordant conclusions (
9). The efficacy of epinephrine as an ESPB adjuvant has been investigated only to a limited extent. Mogahed and colleagues compared ESPB with modified thoracolumbar interfascial plane block using 20 mL of 0.5% ropivacaine mixed with 2 μg/mL adrenaline in lumbar spine surgery. They reported that both techniques reduced postoperative fentanyl requirements, with comparable pain scores over 24 hours, but did not include a plain ropivacaine control group, precluding direct assessment of the epinephrine effect (
21). Our study contributes a direct comparison of ropivacaine with and without epinephrine in an otherwise identical protocol. Not all trials have shown benefit: van de Wijgert et al. found no significant difference in 1-hour pain (NRS 3.8 ± 3 vs 4.2 ± 2.6, P = 0.56), 12-hour opioid consumption (median 11.3 vs 12.5 mg), or time to first opioid (64 vs 41 min, P = 0.09) when comparing bilateral ESPB with ropivacaine alone versus saline in lumbar fusion (
15). In that trial, the block was performed postoperatively and without epinephrine, whereas ours was performed preoperatively and with epinephrine; timing and adjuvant use may both explain the divergent results. Comparable adjuvant data are available for dexmedetomidine, which prolonged block duration and time to first rescue analgesia in ESPB for spine surgery (
25), and for dexamethasone or dexmedetomidine, which extended time to first opioid from 6.2 ± 0.8 h to 12.7 ± 2.7 h and 15.5 ± 2.0 h, respectively (
26). Our median time of 18 h is comparable to the best of these adjuvant results. A Cochrane review of 69 RCTs, including 64 in the meta-analysis, confirmed benefit but noted marked protocol heterogeneity, including frequent adjuvant use, which limits pooled inference for any single adjuvant (
27). Our study addresses this gap with a standardised head-to-head comparison.
Epinephrine has been characterised as an adjuvant in perineural blocks, where local vasoconstriction reduces systemic absorption of the local anaesthetic and prolongs its presence at the target nerve (
16,
17). The erector spinae plane block is an interfascial plane block rather than a true perineural block: the injectate spreads between the erector spinae muscle and the transverse processes over a large, highly vascular surface, and analgesia depends on spread to the dorsal and ventral rami rather than delivery to a discrete nerve. The pharmacokinetics of vasoconstrictor adjuvants may therefore differ, and the dose–response relationship established for peripheral nerve blocks cannot be assumed to apply directly to ESPB. Pharmacokinetic modelling of levobupivacaine with and without epinephrine in transversus abdominis plane and erector spinae plane blocks indicates that epinephrine lowers peak plasma concentration and prolongs absorption in these compartments (
28), supporting a vasoconstrictive mechanism in interfascial planes. However, the magnitude of this effect and its translation into sensory block prolongation have not been established for lumbar ESPB. Our observation is consistent with this mechanism, but we did not measure plasma ropivacaine concentrations or sensory block duration; therefore, extrapolation from perineural to interfascial pharmacokinetics remains inferential.
The higher pain scores at 12 hours in the intervention group (median NRS 5 vs 3; P < 0.0001; adjusted β = 1.48, 95% CI 0.144 - 2.815, P = 0.031) are clinically important and complicate, rather than simply qualify, the interpretation of the early analgesic benefit. We originally proposed that this might represent rebound pain after resolution of a prolonged block. Rebound pain is a well-described phenomenon after long-acting peripheral nerve blocks (
21,
25), and the 12- to 16-hour duration of ropivacaine with epinephrine makes this biologically plausible. However, this explanation was not directly tested: sensory block duration was not measured, rebound pain was not systematically assessed, and no prespecified definition was applied. Rebound pain is therefore one possible explanation among several. Control patients received their first opioid substantially earlier (median 0 h vs 18 h), and this earlier and more frequent analgesic exposure may have provided better pain control at 12 hours, whereas the intervention group may have had less analgesic coverage as the block waned. In the control group, the shorter block may also have allowed a more gradual return of pain, smoothed by earlier oral analgesia (
29,
30). However, the number of patients receiving rescue analgesia before the 12-hour assessment, the cumulative number of rescue doses, and the timing of rescue relative to the 12-hour score were not recorded, so the contribution of differential analgesic exposure cannot be quantified. Differences in surgical procedure mix and complexity, as well as inter-individual variability in block duration and pain trajectories, may also have contributed, and the absence of cumulative opioid data precludes distinguishing a true rebound phenomenon from inadequate analgesic coverage. This finding is exploratory; regardless of mechanism, it indicates that the early analgesic benefit must be weighed against worse pain at 12 hours, and the net clinical benefit remains uncertain.
Several secondary outcomes were tested without adjustment for multiplicity, and multiple regression models were fitted, increasing the risk of false-positive findings. The secondary findings are therefore described as supportive or exploratory and are not used as independent confirmation of a broad analgesic benefit; the primary confirmatory outcome remains PACU pain, and even that result was modest and of uncertain clinical importance. For intraoperative fentanyl, the group medians were identical (50.0 μg), yet the Mann–Whitney test was significant (P = 0.0052), which is possible because the test compares rank distributions rather than medians alone. The Hodges–Lehmann estimate of 50.00 (95% CI 0.00 to 50.00 after rounding; the unrounded lower bound was a small positive number) reflects a shift in the rank distribution, not a difference in medians. Time to first opioid request was analyzed as a continuous outcome, and censoring was not modeled; a formal time-to-event analysis was not performed.
No clinically apparent local anesthetic systemic toxicity, vascular puncture, pneumothorax, or block-related complications were observed. However, given the vasoconstrictive and cardiovascular effects of epinephrine, the absence of observed complications in 61 patients is insufficient to establish comprehensive safety. Detailed hemodynamic changes attributable to epinephrine were not systematically recorded, and episodes of hypertension, tachycardia, arrhythmia, or other potentially epinephrine-related events were not prospectively defined or adjudicated, so broad safety claims should be avoided. Patients with significant cardiac disease were excluded, so the findings cannot be extrapolated to patients with cardiovascular comorbidities, who may be at greater risk of hypertension, tachycardia, or arrhythmia. The total bilateral dose of 200 μg is modest and consistent with doses used in regional anesthesia (
16,
17,
31), but systemic absorption from two injection sites may be additive; plasma epinephrine concentrations and hemodynamic responses were not measured. Future studies should record hemodynamic parameters at standardized intervals before and after block performance, prospectively define and adjudicate epinephrine-related adverse events, and include patients with controlled cardiovascular comorbidity.
5.1. Clinical Implications
First, adding epinephrine to ropivacaine for ESPB is a simple, low-cost intervention that produced a statistically significant but modest early reduction in pain and prolonged the time to the first opioid request; however, the net clinical benefit is uncertain given the higher pain at 12 hours and the absence of cumulative opioid data. Whether this 1-point reduction is clinically important is uncertain, and the intervention should not yet be recommended as definitively effective on the basis of this primary outcome alone. Second, the lower estimated isoflurane consumption is an exploratory finding that should be interpreted cautiously given the vaporiser-based estimation and the difference in surgical duration; it does not establish faster emergence or improved recovery, which were not measured. Third, the higher pain scores at 12 hours are clinically important and complicate the interpretation of the early benefit; this trial did not test any strategy to prevent or mitigate late pain. Therefore, we cannot recommend a specific preventive approach, although future studies should evaluate whether proactive multimodal analgesia before anticipated block regression reduces late pain. Fourth, epinephrine reduces systemic absorption of ropivacaine, offering an additional potential safety advantage, although this was not directly measured in our study.
5.2. Limitations
Several limitations must be acknowledged. Detailed data on surgical procedures—decompression alone versus fusion and the extent of soft tissue dissection—were not collected; therefore, we could neither compare surgical complexity between groups nor adjust for it. The number of levels and surgery duration are incomplete proxies, and residual confounding from unmeasured aspects of surgical complexity remains. The substantial baseline imbalances in ASA class, number of levels, and surgery duration despite randomisation likely reflect the modest sample size and the absence of stratified randomisation by surgical complexity. The fully adjusted model for the primary outcome was borderline (β = −1.250, 95% CI −2.47 to −0.028; P = 0.045), with a wide confidence interval approaching the null. Thus, the adjusted results provide some reassurance but do not exclude residual confounding; the small sample size relative to the number of covariates also raises concerns about model overfitting.
Propensity score matching and inverse probability weighting were considered but not performed. With 61 patients and six covariates, propensity score models would be unstable, and matching would discard a substantial proportion of the sample, reducing precision without removing bias. Moreover, in a randomised trial, treatment assignment is unconfounded by design; the observed imbalances reflect finite-sample chance rather than confounding by indication, and multivariable regression was the appropriate prespecified adjustment. Stratified randomisation in future trials would prevent such imbalances.
Intraoperative fentanyl was analysed as total dose (μg) only; dose per kilogram and dose per hour of surgery were not calculated. Isoflurane consumption was estimated from vaporiser dial settings and duration rather than measured directly with a gas analyser. This method is imprecise, assumes constant fresh gas flow and vaporiser output, and does not account for temperature and pressure effects, circuit losses, or residual agent at the end of surgery. The exact conversion method was not documented in sufficient detail to permit full independent replication. The significant difference in surgery duration (mean 232.3 min vs 161.3 min) may have influenced total fentanyl dose and total isoflurane consumption independently of the intervention effect; normalised analyses (μg/kg/h, mL/h) were not performed. The observed reductions in anaesthetic requirements should therefore be interpreted as approximate and exploratory and are not robust enough, on their own, to support a clinical benefit beyond the primary pain outcome.
The study was not designed to assess rebound pain. Sensory block duration was not measured, rebound pain was not systematically assessed, and cumulative opioid consumption at 12 and 24 hours was not recorded. In addition, the number of patients receiving rescue analgesia before the 12-hour assessment, the exact timing of rescue opioid administration relative to the 12-hour pain score, and whether the 12-hour NRS was obtained before or after the scheduled paracetamol dose were not systematically recorded. Therefore, we cannot determine whether the higher 12-hour pain scores reflect rebound pain, differential exposure to rescue analgesia, differences in surgical complexity, or other factors; this finding should be considered exploratory.
Cumulative postoperative opioid consumption at 12 and 24 hours was not measured. The only postoperative opioid-related outcome was time to first opioid request, which does not capture the total amount of opioid administered after surgery; moreover, patients in the intervention group may have required substantial rescue analgesia after the block resolved. Thus, the findings do not establish a reduction in total postoperative opioid exposure, and the term “opioid-sparing” should be restricted to the intraoperative period. A standardized ward-level rescue analgesic protocol after PACU discharge was not specified for this trial, and the exact agents, doses, and triggers used on the ward were not systematically recorded. Future trials should include cumulative opioid consumption at 12, 24, and 48 hours as key secondary outcomes.
The post-randomization exclusion of 5 patients (7.6% of randomized participants) is a potential source of bias. Although the exclusions were determined before outcome data were collected and were not influenced by treatment effects, excluding participants after randomization can introduce bias. Therefore, the analysis is a modified intention-to-treat analysis, not a full intention-to-treat analysis. The exclusion of patients receiving nonstandard induction agents is particularly relevant because the standardized anesthetic regimen was essential for a valid comparison of intraoperative fentanyl consumption and PACU pain; however, this exclusion may limit generalizability. A sensitivity analysis including all randomized participants was not feasible because the excluded patients had no outcome measurements. Future trials should minimize post-randomization exclusions and prespecify sensitivity analyses, such as worst-case or multiple-imputation analyses.
Several secondary outcomes were tested without adjustment for multiplicity, and multiple regression models were fitted, increasing the risk of false-positive findings. The secondary findings are exploratory rather than confirmatory, and their convergence should not be interpreted as independent confirmation of efficacy. Future studies should prespecify a multiplicity-control strategy.
Time to first opioid request was analyzed using linear regression, which treats the outcome as continuous and does not account for censoring. The observation period was 24 hours, and a formal time-to-event analysis was not performed. If any patients did not request opioids within the observation period, the linear model would not distinguish them from patients who first requested opioids at 24 hours. Whether any patients were censored at 24 hours was not recorded as a distinct variable. Future studies should use formal time-to-event methods.
The safety assessment has several limitations. Although standard intraoperative monitoring (electrocardiography, invasive and noninvasive arterial blood pressure, pulse oximetry, and capnography) was used in all patients, heart rate and blood pressure were not recorded at standardized, protocol-specified intervals specifically before and after block performance to attribute hemodynamic changes to epinephrine. Episodes of hypertension, tachycardia, arrhythmia, or other potentially epinephrine-related adverse events were not prospectively defined, systematically recorded, or adjudicated. Safety monitoring was limited to the perioperative period, and long-term outcomes were not assessed. Therefore, the absence of observed major complications in 61 patients is insufficient to establish the cardiovascular safety profile of epinephrine in this context. In addition, patients with significant cardiac disease were excluded, so the findings cannot be extrapolated to patients with cardiovascular comorbidity. Broad safety claims should be avoided.
The optimal concentration and total dose of epinephrine for ESPB remain uncertain. We studied a single concentration (5 μg·mL-1) at a fixed volume of 20 mL per side, yielding 100 μg per side and 200 μg bilaterally. We did not compare concentrations or doses and therefore cannot determine whether a lower concentration (e.g., 2.5 μg·mL-1) would provide similar prolongation with fewer hemodynamic effects, or whether a higher concentration would provide greater benefit. The dose was extrapolated from peripheral nerve block practice, and the pharmacokinetics of vasoconstrictors in interfascial planes may differ. Plasma ropivacaine and epinephrine concentrations were not measured, and the magnitude of systemic epinephrine absorption from bilateral ESPB is unknown.
Additionally, the sample size (n = 61) was modest, although the study was powered to detect a 1.5-point difference in NRS scores. Post hoc power was not calculated because observed power is a deterministic function of the observed P value and adds no information beyond the confidence interval. We did not measure plasma ropivacaine concentrations to assess whether epinephrine reduced systemic absorption, although the pharmacokinetic effects of epinephrine in ESPB have been characterized previously (
28). Because the block was performed under general anesthesia, formal sensory assessment of block success was not feasible; the lower intraoperative opioid and anesthetic requirements and the prolonged time to first opioid request provide indirect evidence of successful blockade. The exact timing of blinding-guess recording relative to database lock was not documented prospectively. The study was single-center, which may limit generalizability, and data on patient satisfaction, functional recovery, or hospital length of stay were not collected.
5.3. Future Directions
Several avenues for further research emerge. First, future trials should directly measure sensory block duration (e.g., using pinprick or cold sensation testing) to confirm the prolongation inferred from clinical outcomes and use stratified randomisation by surgical complexity to avoid baseline imbalance. Second, longer follow-up (48 - 72 h) should characterise rebound pain after adjuvant-augmented blocks and assess whether pre-emptive multimodal analgesia before block waning mitigates it. Third, given epinephrine’s vasoconstrictive properties, haemodynamic parameters should be recorded systematically at standardised intervals, and patients with controlled cardiovascular comorbidity should be included. Fourth, head-to-head randomised comparisons of epinephrine with dexamethasone and dexmedetomidine are needed to identify the optimal adjuvant for ESPB in spine surgery. Fifth, trials should incorporate patient-reported outcomes, functional recovery metrics, and economic analyses. Sixth, dose-finding studies should define the optimal epinephrine concentration for ESPB. Multicentre trials with larger samples, cumulative opioid endpoints, direct gas analysis, and standardised protocols would improve generalisability and enable robust subgroup analyses.
5.4. Conclusions
In this trial, adding epinephrine to ropivacaine for bilateral ESPB in lumbar spine surgery produced a statistically significant but clinically modest reduction in PACU pain that fell below the prespecified MCID; therefore, the trial was negative for a clinically important primary endpoint. Exploratory secondary analyses suggested lower intraoperative fentanyl use and estimated isoflurane consumption, as well as a longer time to first opioid request; however, these outcomes were not adjusted for multiplicity, isoflurane was estimated without gas analysis, and cumulative postoperative opioid consumption was not measured. Pain was higher at 12 hours in the intervention group, a finding that complicates the early benefit and was not explained by the available data. The analysis followed a modified intention-to-treat principle (five post-randomisation exclusions); detailed surgical procedure data were not collected, and the safety and optimal dose of epinephrine were not comprehensively assessed. The intervention uses a simple, inexpensive, and widely available adjuvant, but its safety profile—particularly regarding epinephrine-related haemodynamic effects—requires further investigation. Larger confirmatory trials with stratified randomisation, cumulative opioid endpoints, direct gas analysis, and systematic haemodynamic monitoring are required before routine adoption can be recommended.