• Anesthesia, Local
• Nerve Block
• Pain, Postoperative
• Acute Pain
• Brachial Plexus

We read ‘Comparing bupivacaine alone to liposomal bupivacaine plus bupivacaine in interscalene blocks for total should arthroplasty: a randomized, non-inferiority trial’ by Elmer et al with interest.1 The authors concluded that non-inferiority was not demonstrated with bupivacaine alone versus liposomal bupivacaine (LB) plus bupivacaine with regard to cumulative opioid consumption over 72 hours after surgery, but that the difference between groups was not meaningful. While we commend the authors for evaluating these outcomes, there are major issues with the study.

First, the authors selected a non-inferiority margin (NIM) of a 22.5 mg difference in opioid consumption in oral morphine equivalents (MMEs) on the basis of previous non-inferiority studies ‘to ensure a clinically relevant difference’.2 3 In the context of the authors’ anticipated opioid utilization rate over 72 hours (45.6 MMEs), this amounts to an ~50% difference in opioid consumption. This assumption reflects a much more stringent definition than previously reported thresholds for opioid consumption after ambulatory surgery.4 Notably, use of large NIMs bias findings in favor of non-inferiority.5 The large NIM was further confounded by use of multimodal analgesia, which resulted in low 72-hour opioid consumption (geometric mean of 19.3 and 31.1 MMEs in the bupivacaine and LB plus bupivacaine groups, respectively).

Second, this study is statistically underpowered, as acknowledged in the Discussion. The authors drew conclusions purely based on point estimates, which is against principles described in Basic statistics for clinicians, which states CIs instead of point estimates should be used when interpreting results (specifically, ‘if clinically important benefits fall within the CI, the trial has not ruled out the possibility that the treatment is worthwhile’).6 Even after considering the proposed 22.5-MME threshold as large and questionable, it resides within the reported CI of improvements with LB plus bupivacaine over bupivacaine (−6.9 to 30.8 MME), and the interval suggests that LB may provide improvements over bupivacaine as large as 30.8 MMEs. Ultimately, the design and interpretation of this study was biased to produce a non-inferior outcome with bupivacaine. It is therefore striking and clinically noteworthy that despite the large NIM and low overall opioid consumption, bupivacaine still did not demonstrate non-inferiority compared with LB plus bupivacaine.

Lastly and of great clinical relevance, the authors failed to contextualize the relationship between pain scores and opioid consumption. Opioid consumption and pain scores are not independent variables; in a study in which patients are permitted opioids for breakthrough pain, one assumes that differences in pain scores may be attenuated over time because patients receiving the less-effective intervention would need greater opioid consumption to manage pain. However, in the current study, pain scores were consistently numerically higher at 24, 48 and 72 hours after surgery in the bupivacaine group compared with the LB plus bupivacaine group (table 3 in Elmer et al), despite the bupivacaine group having not demonstrated non-inferiority compared with the LB plus bupivacaine group for opioid consumption. These results further support the analgesic benefits of LB plus bupivacaine. However, the authors did not account for these observations in the Discussion or their approach to analysis.

In conclusion, the study reported by Elmer et al was designed and interpreted in a manner that biased findings in favor of non-inferiority with bupivacaine compared with LB plus bupivacaine for interscalene blocks in total shoulder surgery. Nonetheless, the data instead demonstrated that interscalene nerve blocks with LB plus bupivacaine provided enhanced postsurgical analgesia over 72 hours, including reduced opioid consumption accompanied by numerical reductions in pain scores. We respectfully request a correction to the conclusions drawn from this study.