Article Text

Impact of ultrasound-guided erector spinae plane block on outcomes after lumbar spinal fusion: a retrospective propensity score matched study of 242 patients
  1. Ellen M Soffin1,
  2. Ichiro Okano2,
  3. Lisa Oezel2,3,
  4. Artine Arzani2,
  5. Andrew A Sama2,
  6. Frank P Cammisa2,
  7. Federico P Girardi2 and
  8. Alexander P Hughes2
  1. 1 Department of Anesthesiology, Critical Care and Pain Management, Hospital for Special Surgery, New York, New York, USA
  2. 2 Orthopaedic Surgery, Spine Care Institute, Hospital for Special Surgery, New York, New York, USA
  3. 3 Department of Orthopaedic and Trauma Surgery, University Hospital Duesseldorf, Duesseldorf, Germany
  1. Correspondence to Dr Ellen M Soffin, Department of Anesthesiology, Critical Care and Pain Management, Hospital for Special Surgery, New York, NY 10021, USA; soffine{at}


Background We evaluated the impact of bilateral ultrasound-guided erector spinae plane blocks on pain and opioid-related outcomes within a standardized care pathway for lumbar fusion.

Methods A retrospective propensity score matched cohort study. Clinical data were extracted from the electronic medical records of patients who underwent lumbar fusion (January 2019–July 2020). Propensity score matching based on common confounders was used to match patients who received or did not receive blocks in a 1:1 ratio. Primary outcomes were Numeric Rating Scale pain scores (0–10) and opioid consumption (morphine equivalent dose) in the first 24 hours after surgery (median (IQR)). Secondary outcomes included length of stay and opioid-related side effects.

Results Of 1846 patients identified, 242 were matched and analyzed. Total 24-hour opioid consumption was significantly lower in the erector spinae plane block group (30 mg (0, 144); without-blocks: 45 mg (0, 225); p=0.03). There were no significant differences in pain scores in the postanesthesia care unit (with blocks: 4 (0, 9); without blocks: 4 (0,8); p=0.984) or on the nursing floor (with blocks: 4 (0,8); without blocks: 4 (0,8); p=0.134). Total length of stay was 5 hours shorter in the block group (76 hours (21, 411); without blocks: 81 (25, 268); p=0.001). Fewer patients who received blocks required postoperative antiemetic administration (with blocks: n=77 (64%); without blocks: n=97 (80%); p=0.006).

Conclusions Erector spinae plane blocks were associated with clinically irrelevant reductions in 24-hour opioid consumption and no improvement in pain scores after lumbar fusion. The routine use of these blocks in the setting of a comprehensive care pathway for lumbar fusion may not be warranted.

  • anesthesia
  • local
  • nerve block
  • pain
  • postoperative
  • ultrasonography
  • regional anesthesia

Data availability statement

Data are available on reasonable request.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Multimodal analgesia and pathway-based care are both advocated as effective approaches for pain management after spine surgery.1 2 However, the optimal regimen has not yet been determined.3 Further, in contrast to other orthopedic surgery subtypes, multimodal regimens for spine surgery which include field and fascial plane blocks are only recently emerging in the literature, and the procedure-specific value and indications have yet to be fully characterized.4 5

To date, one of the most-investigated regional techniques for spine surgery is the erector spinae plane block (ESPB).6 Several prospective studies suggest ESPB provides effective, opioid-sparing analgesia after lumbar decompression and fusion.7–10 Although promising, recent systematic reviews conclude overall low-quality evidence with high risk of bias among studies of ESPB for spine surgery.11 12 Significant heterogeneity in spine surgery subtype, block technique, composition and timing of performance hampers attempts to characterize the benefits of ESPB over conventional analgesia after ‘spine surgery’. Well-designed retrospective studies on this topic are likewise limited and interpretation of these studies is critiqued for small sample sizes and inevitable selection bias.13–16 Thus, evidence from larger samples of patients is warranted, along with assessments of procedure-specific benefits of ESPB for spine surgery.

Accordingly, we conducted a single-center, retrospective propensity score (PS) matched cohort study to assess clinical effects of ESPB on outcomes after lumbar spinal fusion. The primary objective was to compare pain and opioid consumption in the first 24 hours after surgery in patients who underwent lumbar spinal fusion under general anesthesia (GA) with or without bilateral ultrasound-guided ESPB. The secondary objective was to assess differences in length of stay (LOS) and opioid-related side effects between the groups. We hypothesized that there would be significant opioid-sparing effects of ESPB after lumbar spine fusion, that would not be associated with increased pain scores or extended LOS.


Study design and patients

This is a single-center, retrospective, PS matched cohort study.

The electronic medical records (EMR) of all patients who underwent posterior lumbar spinal fusion (with or without decompression) under GA between January 2019 and July 2020 were reviewed to identify cases with the following criteria: age >18 years; primary posterior procedure; received either preincision bilateral ultrasound-guided ESPB or no ESPBs; and elective surgery. Patients with any of the following factors were excluded from analysis: a history of any spine surgery; emergent or revision surgery; procedures via lateral, anterior or combined anterior-posterior approaches; known or suspected spinal infection; any other peripheral, field or fascial plane block, continuous catheter or neuraxial analgesia; unilateral ESPB.

Anesthetic care

We introduced bilateral ultrasound-guided ESPB to spine surgery care at our institution in 2017. However, because of limited evidence to support efficacy, the block has not been formally incorporated into our pathway of care for lumbar spine fusion. Rather, the block is performed at the discretion of the anesthetic-surgical care teams. All patients presenting for lumbar fusion receive standardized perioperative care under an enhanced recovery pathway, as previously described.1 In brief, all patients received GA with endotracheal intubation and a total intravenous anesthetic-based regimen (including propofol, dexmedetomidine and ketamine, but excluding intravenous lidocaine when other peripheral/fascial plane blocks are performed). Standardized, multimodal analgesia including acetaminophen, ketorolac and oral opioids (titrated to pain scores) were provided. Patients were prescribed hydromorphone intravenous patient-controlled analgesia (PCA) for postoperative care on an as-needed basis.

Erector spinae plane blocks

Bilateral ultrasound-guided ESPBs were performed with the patient in prone position after induction of GA, prior to surgical incision. A C60 curved array ultrasound probe (FUJIFILM Sonosite, Washington, USA) was placed in parasagittal orientation in the midline to identify the spinous processes. The probe was translated laterally until the tips of the transverse processes were viewed. A 20-Ga 4-inch Ultraplex needle (B. Braun Medica, Pennsylvania, USA) was placed in-plane and advanced in a cranial-to-caudal direction until the tip was under the erector spinae plane. Depending on patient body mass index (BMI), between 20 and 30 mL 0.25%–0.375% bupivacaine with or without adjuvant (preservative-free dexamethasone, 2 mg/30 mL bupivacaine) was injected bilaterally at the tips of the transverse process/under visual guidance. The choice and volume of local anesthetic was at the discretion of the individual anesthesiologists.

Data collection

Data were retrospectively extracted from the EMR. Demographic variables included age, sex, BMI, race, American Society of Anesthesiologists (ASA) classification, smoking status, history of anxiety and/or depression, and preoperative opioid tolerance (opioid use on most days for >3 months).17 Perioperative variables included: the quantity of intraoperative opioid administered, indication for surgery, number and anatomic levels fused, any additional surgical interventions (lumbar decompression; number and specified levels), and duration of surgery. Postoperative variables collected included: Numeric Rating Scale (NRS) pain scores, opioid consumption, opioid-related side effects and hospital LOS.

NRS pain scores are assessed on a scale of 0–10 by nursing colleagues as part of our institutional standard of care and entered in the EMR. NRS was assessed hourly while patients were in the postanesthesia care unit (PACU) and 6-hourly on the nursing floor. Scores reflect the pain at the time the assessment is performed. Details of resting versus movement scores are not recorded. Opioid consumption was quantified according to type, dose, frequency, and duration of use. Using standard conversion values, equianalgesic doses were calculated for different types of opioids based on morphine equivalent dose (MED), expressed as milligrams of morphine.18 LOS was measured continuously and expressed as number of days. Opioid-related side effects were defined a priori as postoperative nausea/vomiting (indicated by administration of rescue antiemetic medications), sedation/respiratory depression (indicated by escalation of care and/or need for supplemental oxygen and/or naloxone), pruritus (indicated by administration of nalbuphine or diphenhydramine), constipation/obstipation, and cognitive/central nervous system changes (dizziness, confusion, and disorientation).


The primary outcome was total opioid consumption and NRS pain scores during the first 24 hours after surgery. Total opioid consumption comprised the sum of oral and intravenous (clinician-administered and intravenous PCA) opioid used, during the intraoperative, PACU, and floor phases of care.

Secondary outcomes included differences between the groups in: opioid-related side effects, LOS, duration of intravenous PCA use, opioid-consumption between 24 hours and discharge, and quantity of opioid prescribed at discharge from the hospital.

Statistical analyses

Given lack of baseline efficacy data, no statistical power calculation was conducted before the study. The sample size was solely based on the available data from consecutive patients meeting criteria for inclusion in analysis. A minimum clinically meaningful effect size was likewise not defined a priori. Outcomes with missing data were analyzed by last observation carried forward (NRS) or not analyzed further (LOS).

Normality of data distribution was assessed using the Shapiro-Wilk test. Continuous variables are summarized as medians (range). Categorical variables are summarized as counts (%). Baseline characteristics between the ESPB and no-ESPB groups were summarized and compared using bivariable tests, Mann-Whitney U for continuous variables and Fisher’s exact tests for categorical variables, respectively.

The rationale and methods for PS matching to evaluate cause-and-effect relationships in retrospective studies have been well described.19 Briefly, the PS was calculated by logistic regression analysis using patient demographics and baseline clinical data to examine the association of ESPB or no-ESPB with outcomes of interest. The independent variable was bilateral ultrasound-guided ESPBs. The PS represented the estimated probability of receiving an ESPB. The following covariates with potentially relevant effects on the study outcomes were included in the model, due to their potential influence on outcomes of interest: age, BMI, race, ASA classification, smoking status, history of anxiety, depression, chronic pain condition, preoperative opioid use, preoperative gabapentin or pregabalin use, preoperative diagnosis, number of lumbar levels fused/decompressed, and surgical level.

Patients in the ESPB and no-ESPB groups were matched using the PS in a 1:1 ratio. The balance of baseline covariates between groups was deemed acceptable when the standardized mean difference (SMD) (the absolute difference of the group means as a percentage of their pooled SD) was <10% for most variables. Unpaired cases were excluded from analysis. Statistically significant differences between the groups were defined using two-tailed tests, where p<0.05. We additionally performed a sensitivity analysis by calculating the E-values for any outcome variable with remaining statistical significance after PS matching. The E-value is a measure which indicates the robustness of unmeasured confounders. A large E-value implies that substantial unmeasured confounding is necessary to explain observed significance differences.20 All statistical analyses were performed using R software (V.4.0.3).


Demographics and perioperative variables

The initial search of the EMR retrieved 1846 lumbar spine surgeries between January 2019 and July 2020. After eliminating records according to eligibility criteria (n=1476), 370 remained. Of these, 170 patients had surgery with ESPB and 200 had surgery with no-ESPB. After PS matching, 242 patients were included in the final analysis (121 patients per group; figure 1).

Figure 1

Flow of cases through selection for analysis. ESP, erector spinae plane; ESPB, erector spinae plane block.

The demographic characteristics and perioperative data for the total study cohort (unadjusted analysis) are shown in table 1. Significant statistical differences were found for age, indication for surgery and location of levels fused. Table 2 shows the group characteristics after PS matching. There was reasonable balance on demographics for most variables with slight imbalances found for preoperative gabapentin/pregabalin use (SMD 0.18) and the number of patients who underwent additional decompression at the time of fusion (SMD 0.18) (table 2).

Table 1

Patient demographics and unadjusted comparisons

Table 2

Patient demographics after Propensity Score matching

Online supplemental table 1 shows properties of the ESPB composition and volume. All blocks were performed using bupivacaine, with either 0.25% (n=38 patients), 0.375% (n=67 patients), or 0.5% (n=16 patients). Of these, 46 (38% of patients) pairs of blocks included preservative-free dexamethasone. Total volume ranged from 20 to 30 mL per side, per patient (online supplemental table 1).

Supplemental material

Primary outcome (opioid consumption and NRS pain scores in the first 24 hours after surgery)

Total opioid consumption in the first 24 hours after surgery was statistically significantly lower in the ESPB group compared with the no-ESPB group on both the unadjusted comparisons (ESPB: 25 mg MED (0, 144); no-ESPB: 45 mg MED (0, 225 mg); p<0.001) and after PS matching (ESPB: 30 mg MED (0, 144); no-ESPB: 45 mg MED (0, 225); p=0.03) (table 3). Total intravenous PCA opioid consumption was also significantly lower in the unadjusted (ESPB: 4 mg MED (0, 89); no ESPB: 12.5 mg (0, 220); p<0.001) and PS matched comparisons (ESPB: 6 mg MED (0, 89); no-ESPB: 12 mg MED (0, 220); p<0.001).

Table 3

Unadjusted and Propensity Score (PS) matched comparisons: primary and secondary outcomes

There were no significant differences in NRS pain scores in the PACU between the groups on either the unadjusted (ESPB: 4 (0, 9); no-ESPB: 4 (0,8); p=0.122) or PS matched comparisons (ESPB: 4 (0, 9); no-ESPB: 4 (0,8); p=0.984) (table 3). Likewise, there were no differences in NRS pain scores on the nursing floor in either the unadjusted (ESPB: 4 (0,8); no-ESPB: 4 (0,8); p=0.342) or after PS matching (ESPB: 4 (0,8); no-ESPB: 4 (0,8); p=0.134). Pain scores were missing from the EMR in 22 patients who received ESPB (18%) and 13 patients who did not (11%).

The calculated E-values of statistically significant outcomes were 1.30–1.39 (online supplemental table 2).

Supplemental material

Secondary outcomes

Total LOS was 6 hours shorter in the ESPB group on unadjusted comparison (ESPB: 75 hours (9, 411); no-ESPB: 81 (19, 287); p<0.001) and 5 hours shorter after PS matching (ESPB: 76 hours (21, 411); no-ESPB: 81 (25, 268); p=0.001). There were no significant differences in PACU LOS between the groups in either analysis, but LOS on the nursing floor was shorter in unadjusted (ESPB: 61 hours (8, 399); no-ESPB: 69 hours (16, 279); p<0.001) and after PS matching (ESPB: 64 hours (8, 399); no-ESPB: 69 hours (16, 260); p=0.001) (table 3). Eight patients who received ESPBs (7%) and three who did not (3%) were discharged from the hospital before 24 hours.

Duration of intravenous PCA use was statistically significantly shorter in the unadjusted analysis (ESPB: 22 hours (3.6, 53); no-ESPB: 24 hours (0, 235); p=0.005) but not after PS matching (ESPB: 22 hours (5, 53); no-ESPB: 25 hours (0, 235); p=0.098) (table 3).

Significantly fewer patients who received ESPB required postoperative antiemetic administration in both the unadjusted comparison (ESPB: n=107 (63%); no-ESPB: n=167 (84%); p<0.001) and after PS matching (ESPB: n=77 (64%); no-ESPB: n=97 (80%); p=0.006) (table 4). There were no other significant differences in opioid-related side effects or outcomes between the groups.

Table 4

Unadjusted and Propensity Score (PS) matched comparisons: complications and opioid-related side effects


In this retrospective PS matched cohort study of 242 patients undergoing lumbar fusion, ESPB was associated with minor reductions in 24-hour opioid consumption, antiemetic use and LOS without any significant effects on NRS pain scores at any phase of care. These results are consistent with the growing body of literature describing the beneficial impact of ESPBs in spine surgery cohorts. However, the limited effects described here suggest the routine addition of ESPB to comprehensive pathway-based care for lumbar spine fusion may not be warranted, although could be considered on a case-by-case basis.

Although the association between ESPB and early opioid-sparing effects was statistically significant, this is of uncertain clinical importance. We found a total median difference of 15 mg MED over 24 hours, which equates to 10 mg oxycodone. In comparison to other studies of ESPB for lumbar surgery, the opioid-sparing effect was less impressive.21–25 A possible explanation is that we included comprehensive multimodal analgesia in our pathway, together with patient education and expectation-setting regarding the judicious use of opioids in recovery.1 There are few reports of pathway-based care for spine surgery which also incorporate regional techniques, and the effects of the pathway itself on opioid consumption should be considered in parallel with the effects of ESPB. For example, although we found that total intravenous PCA opioid consumption was statistically lower among patients who received ESPB, the duration of use was not different between the groups. This is likely to reflect the use of standardized order sets, in which a preset time for removing the intravenous PCA is integrated into the care pathway.1

Patients receiving ESPB had a modestly shorter LOS, driven by the LOS on the nursing floor. Thus far, LOS has been inconsistently assessed in studies of ESPB for spine surgery. Further, effects are equivocal, with some15 16 but not all studies8 24 concluding that ESPB reduces LOS after lumbar spine surgery. As for other outcomes found here, the difference of 5 hours that we associated with ESPB may not represent a clinically significant advance on an individual patient basis. However, as demand for lumbar surgery continues to rise, a 5-hour saving in LOS could result in significant gains when extrapolated to hospitals or healthcare systems as a whole.

The opioid-sparing effects of ESPB were accompanied by a presumptive reduction in PONV, as assessed by administration of anti-emetic medications. PONV is a key driver of patient satisfaction, recovery and readiness-for-discharge.26 Thus, lower incidence of PONV could have contributed to the shorter LOS found ESPB group. Several studies have measured the incidence of PONV after ESPB for lumbar surgery, and all conclude lower rates among patients who received blocks.7–9 21 Interestingly, we found the overall incidence of antiemetic use was high in both groups and higher than expected in the ESPB group based on published reports of PONV benefits afforded by the block.7–9 21 However, when considering that no differences were found for other opioid-related side effects, and that there was a modest reduction in total opioid consumption, this result is perhaps less surprizing.

A major unexpected finding of our study was that lower opioid consumption over the first 24 hours in the ESPB group was not accompanied by improvements in pain scores at any phase of care. According to the results of three systematic reviews of ESPB for spine surgery, the block has a significant effect on reducing NRS scores up to 24 hours postsurgery, with the most consistent benefits found in the first 4–6 hours.11 12 25 On the other hand, a significant publication bias has been associated with the pain benefits of the block for spine surgery, suggesting the true effect in the population remains to be characterized.25 There are several explanations for the lack of difference in NRS scores described here: First, acute pain after lumbar spine surgery is complex and multifactorial, involving multiple nociceptive and neuropathic pathways. Recent studies have suggested the most plausible mechanism of action of ESPB is via local anesthetic spread to the neural foramina and dorsal root ganglia, however, the extent of the sensory block itself appears to be highly variable.27 It is possible that ESPB does not comprehensively address the multiple sources of pain after lumbar fusion in a consistent fashion. Second, due to limitations of the retrospective design, we were unable to measure block duration or efficacy. We chose 24 hours as our primary endpoint, but this may have been longer than the duration of the ESPB. Third, we analyzed NRS scores extracted from the EMR at predetermined times. This may fail to capture qualitative effects of ESPB on pain, including influence on resting versus movement, or best versus worst pain scores over the early recovery phase. Finally, we did not assess the incidence of ESPB success or failure. If these factors were significant, the benefits of the block would be underestimated in this population. Finally, some degree of heterogeneity in technique and approach may account for the lack of observed effect on pain scores found here, compared with earlier studies.

Strengths and limitations

Strengths of our study include the large sample of PS matched patients, all of whom were cared for under a standardized enhanced recovery pathway for lumbar spine fusion. We included a mixed cohort of opioid-tolerant and naïve patients, thereby increasing the generalizability of the findings. Although we found statistically significant effects of ESPB on several outcomes we interpreted our results in the context of clinically important differences.

There are also several limitations to our study. First, due to the non-randomized design, the results only indicate associations between ESPB and pain and opioid-related outcomes. Second, PS matching will have reduced the risk of confounding and selection biases, but we could only account for confounders we could identify, and for which relevant data were available. Residual bias due to uncontrolled confounding variables remains a possibility. Although the results of the sensitivity analysis suggested this risk is small, unmeasured factors influencing both the choice to perform ESPB and outcomes of interest may have persisted.20 Third, although we consider the enhanced recovery pathway to be a strength of the design, there may have been a ceiling effect, whereby additive analgesic benefits of ESPB could not be revealed, thereby underestimating the effect of the block in lumbar surgery. Fourth, the spinal level at which the ESPBs were performed was at the discretion of the performing provider and was not evaluated here. By convention, our institutional practice is to perform the blocks at the most proximal transverse process of included spinal levels, however, these details are not recorded in the institutional (templated) block note. The impact of any variation in injection site on the results reported here is uncertain, but likely to be small. There is currently no consensus regarding the optimal location for ESPB performance for lumbar spine surgery, with multiple published studies concluding efficacy after injection at a range of locations.8 28–30 Although early reports of successful analgesia after lumbar surgery advocated using a fixed low thoracic injection point,28 29 more recently, this approach has been questioned, based on limited distal lumbar spread of local anesthetic.30 These latter results may be consistent with the minimal benefits found here on pain scores and opioid consumption, and help to define the procedures and indications for ESPB. Finally, the study was conducted at an academic orthopedic surgery specialty hospital, limiting generalizability to other practice settings.


In conclusion, we found that adding bilateral ultrasound-guided ESPBs to a pathway for lumbar spinal fusion was associated with minor reductions in opioid consumption, antiemetic use and LOS, without affecting NRS pain scores. Although statistically significant, the marginal clinical benefits do not support the routine addition of ESPBs to comprehensive lumbar spine fusion care pathways. These results help to define procedure-specific indications and benefits for ESPB. For example, ESPB may show benefits on outcomes after more extensive spine surgery (involving multiple lumbar and/or thoracolumbar levels), with higher anticipated pain burden and systemic inflammatory response. Provocatively, the small clinical effects found here are similar to those described for ESPB in breast surgery31 and warrant further study in prospective, well-designed, adequately powered trials, and in further retrospective studies from a range of practice settings.

Data availability statement

Data are available on reasonable request.

Ethics statements

Patient consent for publication

Ethics approval

The study was conducted in accordance with the ethical principles of the Declaration of Helsinki and approved by the institutional review board of Hospital for Special Surgery (HSS-IRB #2020-1877). Written informed consent was waived.


We are grateful to Jennifer Shue, MS for administrative oversight of this study, and to Zhaorui Wang, Ikenna Onyekwere and Shuting Lu for data extraction and verification of data accuracy.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Twitter @ESoffin

  • EMS and IO contributed equally.

  • Contributors EMS, APH and IO conceived and designed the study; all authors participated in the planning, conduct, reporting, and interpretation of data; LO, AA and IO participated in acquisition of data; IO performed statistical analyses; EMS wrote the manuscript; all authors participated in revising the manuscript for content. All authors approved the final version of the manuscript; EMS accepts full responsibility for the work and conduct of the study, had access to the data, and controlled the decision to publish.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Linked Articles