Free
Correspondence  |   June 2019
Supraclav Suprascap Interscalene Shoulder Surgery: Reply
Article Information
Correspondence
Correspondence   |   June 2019
Supraclav Suprascap Interscalene Shoulder Surgery: Reply
Anesthesiology 6 2019, Vol.130, 1086-1087. doi:https://doi.org/10.1097/ALN.0000000000002714
Anesthesiology 6 2019, Vol.130, 1086-1087. doi:https://doi.org/10.1097/ALN.0000000000002714
In Reply:
We thank Dr. Ganesamoorthi et al. for their interest in our article1  and their commentary regarding our research. We would like to address each of their concerns.
First, the primary outcome revealed analgesic noninferiority of an anterior suprascapular block compared to an interscalene block for rotator cuff shoulder surgery (P = 0.012) in the postanesthesia care unit. In contrast, noninferiority was not shown when comparing supraclavicular to interscalene blocks with the same criteria (P = 0.088). The discussion in our initial publication addresses these findings at length along with our best interpretation for the collected data. Indeed, the injection endpoint for the supraclavicular group was at the superior and middle trunks. However, the exact postinjection distribution of the 15 ml of local anesthetic remains unknown. Clinical care of our patients should be based more on clinically relevant outcomes1,2  rather than conjecture of how local anesthetic anatomically flows around the brachial plexus at various points of injection.
Second, our pain variables did not exceed the expected skew for the normal distribution (conventionally −1 to 1), although the means were higher than the medians, reflecting a few subjects with higher pain scores (table 1). While within-subject changes are typically more normally distributed than individual values, analyzing pain in terms of a change from baseline is not typically performed in our field of study and we have been asked to remove these analyses from previous research. Our methods and plan for the analysis of pain was always a comparison of means and we did not feel that there was sufficient reason to deviate from that plan given the realized data.
Table 1.
Numerical Rating Scale Average and Median Postanesthesia Care Unit Pain Scores
Numerical Rating Scale Average and Median Postanesthesia Care Unit Pain Scores×
Numerical Rating Scale Average and Median Postanesthesia Care Unit Pain Scores
Table 1.
Numerical Rating Scale Average and Median Postanesthesia Care Unit Pain Scores
Numerical Rating Scale Average and Median Postanesthesia Care Unit Pain Scores×
×
The final topic addressed is whether mean pain is an appropriate statistic, as the CIs include scores less than 0 and are impossible values. On the numerical rating scale, scores of 2.1 and 2.6 (the mean average pain scores) are also improbable values for individual patients, as patients rarely respond with such precision when asked for a numerical rating scale pain score of 0 to 10. Studies designed using comparisons of means tend to be more powerful than comparisons based on other statistics, and the mean may be a useful measure—at least mathematically—to compare groups.3–5  However, the mean of the distribution may be less useful for communicating expectations to patients. In this study, half the subjects had scores less than 2 for average pain. While we were pleased with this low pain burden across groups, we agree that it complicates the interpretation of comparisons based on means. When pain score changes are appropriate—for example, the change from 24 to 48 h—the proportion of subjects with reduced pain is a highly useful statistic. Consensus on critical thresholds in the numerical rating scale might allow for comparisons of proportions of subjects with levels of severity of pain. However, the numerical rating pain scale is applied to a wide variety of surgeries and other types of pain, so this consensus may be difficult to reach and is outside the scope of this research. Regardless, these decisions should be made before commencement of the study to be appropriately incorporated into the statistical design.
Competing Interests
The authors declare no competing interests.
References
Auyong, DB, Hanson, NA, Joseph, RS, Schmidt, BE, Slee, AE, Yuan, SC : Comparison of anterior suprascapular, supraclavicular, and interscalene nerve block approaches for major outpatient arthroscopic shoulder surgery: A randomized, double-blind, noninferiority trial. Anesthesiology. 2018; 129:47–57. [Article] [PubMed]
Auyong, DB, Yuan, SC, Choi, DS, Pahang, JA, Slee, AE, Hanson, NA : A double-blind randomized comparison of continuous interscalene, supraclavicular, and suprascapular blocks for total shoulder arthroplasty. Reg Anesth Pain Med. 2017; 42:302–9. [Article] [PubMed]
MacCallum, RC, Zhang, S, Preacher, KJ, Rucker, DD : On the practice of dichotomization of quantitative variables. Psychol Methods. 2002; 7:19–40. [Article] [PubMed]
Royston, P, Altman, DG, Sauerbrei, W : Dichotomizing continuous predictors in multiple regression: A bad idea. Stat Med. 2006; 25:127–41. [Article] [PubMed]
Anderson, JJ : Mean changes versus dichotomous definitions of improvement. Stat Methods Med Res. 2007; 16:7–12. [Article] [PubMed]
Table 1.
Numerical Rating Scale Average and Median Postanesthesia Care Unit Pain Scores
Numerical Rating Scale Average and Median Postanesthesia Care Unit Pain Scores×
Numerical Rating Scale Average and Median Postanesthesia Care Unit Pain Scores
Table 1.
Numerical Rating Scale Average and Median Postanesthesia Care Unit Pain Scores
Numerical Rating Scale Average and Median Postanesthesia Care Unit Pain Scores×
×