LAUSR

In their recent article in Advances in Therapy, Gottschlich et al. [1] concluded that the phytomedicines ELOM-080 and BNO 1016 were well tolerated and were of comparable effectiveness in treating acute rhinosinusitis (ARS). We noted the proposed advantages of treatment with ELOM-080 with great interest, but on examination of the article we identified several weaknesses that bring the validity of the authors’ conclusions into question. Our primary concern was that the exploratory conclusions of a non-interventional study were being inappropriately presented, in the manner of a superiority trial, as statistically and clinically relevant results, whereas non-interventional studies (NISs) are typically used to examine the safety of existing treatments rather than demonstrate the superiority of one treatment over another. Other concerns, which we discuss in more detail below, include a lack of clarity of the role of the study sponsor, apparent bias in the discussion section, inaccuracies in the figures and incorrect references to previously published data, and these points lead us to feel that this article presents a misleading perspective on the results of the study. NISs provide a valuable perspective of treatment in real-world practice, but the methodology described in the article is beyond the scope of a conventional NIS. The methods used were closer to the set-up of a randomised controlled clinical trial and the results are presented as such. However, these results have been presented without the statistical support that would have been expected of a comparative trial, such as predetermined endpoints, controlling for the influence of concomitant medications, and whether the correct adjustments have been made for type 1 error, representative sampling and multiplicity. Randomised controlled trials comparing treatments rely on analyses derived from full analysis sets to draw conclusions on the potential superiority of one treatment versus another, yet in this study, data from the full analysis set appear to be missing. Our other methodological concerns include the inclusion of Likert scale and numeric rating scales to record symptoms, the use of which is typically restricted to clinical trials, and the lack of a meaningful total symptom score. C. Bachert (&) Head Upper Airways Research Laboratory (URL), University Hospital Gent, Ghent, Belgium e-mail: [email protected]