Regional Anesthesia for Knee Surgery: Finding Positives in the Negatives

See Article, page 1163 “To block or not to block…that is the question….” In a non-Shakespearean tone, this seems to be the prevailing thought amongst regional anesthesiologists. With a variety of infiltration-based and perineural regional analgesic techniques, judiciously selecting an appropriate, yet evidence-based strategy, for postoperative analgesia is challenging. Numerous investigations have assessed various regional techniques for total knee arthroplasty (TKA) including femoral nerve block, adductor canal block (ACB), genicular nerve blocks, infiltration between the popliteal artery and posterior capsule of the knee (iPACK), and periarticular local infiltration (LIA).1 The simplest of approaches is for all patients undergoing TKA to receive LIA. In practice, there is a regional anesthetic “à la carte menu” whereby clinicians pick and choose the regimen that they believe has the best evidence. It is noteworthy that interventions can be compared through different methods in randomized trials. In particular, it is important that clinicians understand that there are distinct differences between the superiority design (the “conventional design”) and the noninferiority study design.2 Typically, regional anesthesiologists, want to know if 1 intervention is superior (with a clinically important clinical difference) compared to another. In contrast, a noninferiority design provides leeway and allows us as clinicians to decide whether intervention A is not “worse” than intervention B, and if this finding warrants a change in practice. On a more granular level, 1 key to understanding these types of trial designs is to make note of the a priori noninferiority margin and the 95% confidence interval between the 2 treatments (Figure). In a unidirectional test, demonstrating noninferiority requires (1) the 95% confidence interval for the difference between the interventions compared does not reach a threshold of statistical significance (implying the null hypothesis could not be rejected); and (2) the upper or lower boundary of the 95% confidence interval for the difference excludes the noninferiority margin: the value that is considered clinically important. In regional anesthesia, comparing nerve blocks in the setting of multimodal analgesia may be challenging since detecting a sizable difference in treatment effect while maintaining a manageable sample size for superiority trials is difficult. As an alternative, researchers may resort to equivalence trials; but these too require relatively large sample sizes. As such, the use of a noninferiority trial design is slowly gaining popularity in regional anesthesia research, as it allows designing trials that answer the question “does this intervention provide no-worse analgesic effects than what is typically performed for this surgery?,” while keeping sample sizes manageable.Figure.: Interpretation of superiority and noninferiority study designs. Comparison is between standard of care (S) and alternative treatment (A). Margins of superiority and noninferiority are delineated by δ. *Null hypothesis could not be rejected.In this issue of Anesthesia & Analgesia, YaDeau et al3 conduct a well-designed noninferiority randomized trial assessing the effectiveness of adding LIA to preexisting single-injection ACB and iPACK in patients undergoing TKA under spinal or combined spinal-epidural. Frequently, trials evaluating regional anesthetic therapies for TKA consider LIA to be the “gold standard.”4 Therefore the regional anesthetic is a therapy additional to LIA; however, in this study, the investigators “flip the switch” and assess the utility of adding LIA to preexisting nerve block. The study evaluates the noninferiority using a margin of + 1.0 numeric rating scale (NRS) unit of ACB, iPACK, and placebo LIA versus ACB, iPACK, and active LIA with a local anesthetic. The primary outcome was NRS during activity at 24 hours. The abovementioned margin is reasonable considering that the minimum clinically important difference (MCID) for TKA has been reported as 1.6 to 2.2 NRS units.5 The aims are (1) to demonstrate the lack of statistically significant difference in postoperative pain and (2) to show that the confidence interval excludes the a priori NRS margin. For their primary outcome, YaDeau et al observed a nonsignificant mean difference (reported as placebo LIA minus active LIA) in NRS pain with active LIA (mean difference 0.3 NRS units, 95% confidence interval, -0.9 to +1.5). Using this information, and the above-defined criterion, noninferiority has not been shown between these interventions, as the upper bound of the 95% confidence interval (+1.5 NRS units) included the noninferiority margin of 1.0 NRS unit. In their trial, the authors also observed a nonsignificance between the 2 interventions for other clinically important outcomes, including rest and active pain at short-term time intervals, analgesic consumption at all time points, and quality of recovery. That said, 3 points have attracted our attention and made us wonder if things could have been done differently. First, a noninferiority design should be reserved for the aim of testing if an alternative treatment, with certain advantages, is not worse than the standard of care: in this present study, LIA is combined with blocks. Interestingly, YaDeau et al followed a novel approach in which a placebo comparator was used to test whether “lack of treatment” is not worse than receiving treatment. This is critical to understanding their noninferiority study design, since the mean difference for this trial is in relation to placebo LIA: placebo LIA minus active LIA. The alternative treatment in this situation, placebo LIA, is not purported to provide any additional advantages, thus justifying the experiment. Abandoning a treatment (LIA) whose efficacy has been demonstrated in favor of placebo may be misconstrued to challenge the principles of equipoise, but the authors provide telling justification. Second, the findings may have varying implications on clinical practice. Concluding that the intervention examined (placebo LIA) is not “non-inferior” to active LIA is the accurate interpretation of having a difference in means (95% confidence interval) that crosses both zero and the pre-designated noninferiority margin (1.0 NRS unit). This particular result pattern is statistically referred to as “inconclusive.” With no statistical difference detected in opioid consumption and other secondary analgesic outcomes, this study raises important unanswered questions because the authors could not demonstrate that eliminating LIA yields pain outcomes that are not worse. In practice, clinicians may interpret these findings as grounds either to continue combining LIA with peripheral nerve blocks or to abandon LIA altogether in favor of peripheral blocks. Third, some may argue that LIA constitutes the essential, cornerstone intervention in post-TKA analgesia, and any additional interventions added to LIA yield incremental benefits that are not clinically important.1 Therefore, based on this work, an interesting future trial design would be to examine saline versus active ACB and iPACK in the setting of LIA. The approach in the current study by YaDeau et al seems to suggest that LIA is not the gold standard by attempting to prove that a placebo treatment is not worse than LIA. In failing to demonstrate noninferiority, this well-crafted designed and conducted trial communicates several positive lessons that can be applied to future research investigating the ideal analgesic regimen for TKA. Clearly, the case is yet to be closed! DISCLOSURES Name: Nasir Hussain, MD, MSc. Contribution: This author helped prepare the article and revise the work. Name: Tristan Weaver, MD. Contribution: This author helped prepare the article and revise the work. Name: Faraj Abdallah, MD, MSc. Contribution: This author helped prepare the article and revise the work. This manuscript was handled by: Michael J. Barrington, MB BS, FANZCA, PhD.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.