Reply to Vozoris: Opioids for Dyspnea in Chronic Obstructive Pulmonary Disease: Short on the Details

Abstract

pooled analysis in detail (such as with a Forrest plot) and explain that such detail will be independently published later. Nici and colleagues (1) neglect to specify for the reader that out of the 12 trials included in their meta-analysis on opioids for dyspnea in COPD, only 3 reported statistically significant positive results for opioids over placebo, and the remaining 9 were negative. Furthermore, one of the three positive trials involved individuals with COPD, not secondary to tobacco smoke exposure but instead secondary to mustard gas (2); thus, this study is associated with bias. Along with the overall pooled estimate, it would have been helpful for Nici and colleagues (1) to concurrently present such important details, to provide readers with a more comprehensive and balanced view of their meta-analysis. When considering the results of a metaanalysis, it is instructive to know if a positive signal is being driven by a majority of studies included, versus a small number, and if the latter case, whether such studies might be associated with bias. The authors also overlook acknowledging two other recently publishedmeta-analyses on the topic of opioids for dyspnea in COPD (3, 4), using nearly the same evidence base yet reporting strikingly different findings. Considering 10 out of 12 trials that Nici and colleagues (1) did, Ekström and colleagues (3) in 2015 reported a markedly lower SMD in dyspnea scores for opioids over placebo (20.35; 95% CI, 20.53 to 20.17). Subsequently, in 2016, a meta-analysis was published by Barnes and colleagues (4), and when considering studies involving only individuals with COPD, this group reported an SMD in dyspnea scores similar to that of Ekström and colleagues (5), but not statistically significant (SMD 20.49 [95% CI, 21.08 to 0.10] for trials where dyspnea scores were compared with baseline, and SMD20.21 [95% CI, 20.45 to 0.04] for trials where dyspnea scores were compared with the pretreatment period). The SMD estimates from the aforementioned two meta-analyses show, at best, a small improvement in dyspnea intensity with opioids and fall below the threshold that Nici and colleagues (1) set as clinically meaningful (SMD .0.50). It is challenging to reconcile the SMD estimate of Nici and colleagues (1) with that of Ekström and colleagues (3) and Barnes and colleagues (4), without more details being provided by the former authors. Finally, Nici and colleagues’ (1) literature search terminated in July 2019. However, since then, two more randomized controlled trials have been published that evaluated opioids for dyspnea in advanced COPD (5, 6). Both trials reported negative results, and the study by Currow and colleagues (5) is the largest and, arguably, best-quality trial on the topic conducted to date. Therefore, Nici and colleagues’ (1) recommendation regarding opioids for dyspnea in COPD does not incorporate the most up-to-date, best-quality evidence on the topic. On such an important and controversial topic as using opioids to treat refractory dyspnea in COPD, in a guideline document, it behooves Nici and colleagues (1) to provide readers much more detail about their meta-analysis, including what, why, and how data got pooled. n