Dear Editor
This paper reports on a high-quality network meta-analysis of randomized trials, focusing on several treatments for chronic pain caused by temporomandibular disorders (TMD).
A network meta-analysis allows comparison of the effectiveness of different treatments, even when these treatments have not been compared directly in the primary studies. This also provides a sense of the level of improvement that can be expected in routine practice, if innovation is implemented (i.e., if treatments which have shown to be efficacious or effective, but are not used in clinical practice, become generalised). Bearing in mind the prevalence of TMD and that many patients treated in routine practice report unsatisfactory results, this could be informative and helpful to improve patient management.
Unfortunately, these goals were not fully achieved because the authors of the network meta-analysis on TMD decided to restrict analysis to treatments that are already commonly used in routine practice.
This decision may have reduced the resources required to complete this study, and can be seen as a pragmatic approach. However, it is well known that “availability in routine practice” and “high quality evidence of efficacy or effectiveness” do not always correlate, and prioritising availability may be more appropriate for a clinical guideline destined for implementation in a specific geographical setting, than for a network meta-analysis published in a journal with an international scope.
This decision also introduces a risk of bias, caused by eliminating data from the analysis arbitrarily. In fact, no definition of “common use” is provided, and trials selected for inclusion appear to have been selected based solely on the perception of six clinical experts on the availability of the treatments assessed in those trials, in their own environments (UK, Canada and US, for the experts for which such information is disclosed). As a result, data on evidence-based treatments not routinely used in those settings would be overlooked, even if randomized trials had shown them to be more effective than the treatments commonly used in those environments. This does not contribute to improving treatment for TMD and might reflect circular reasoning, if the underlying undisclosed assumption is that routine treatment in those settings represent the ultimate “gold standard”.
Given the high quality of the search and statistical methods used in this study, as well as the effort and resources it involved, we think that the authors could significantly improve its validity, global reach and clinical usefulness, should they consider to re-analyse the data, including the evidence presented by the 10 studies that the clinical experts chose to disregard.
A high-quality network meta-analysis which could improve clinical usefulness by using the complete available dataset