Peer Review #3 of "Effectiveness and adequacy of blinding in the moderation of pain outcomes: Systematic review and meta-analyses of dry needling trials (v0.1)"

Abstract

Blinding is critical to clinical trials because it allows for separation of specific intervention effects from bias, by equalising all factors between groups except for the proposed mechanism of action. Absent or inadequate blinding in clinical trials has consistently been shown in large meta-analyses to result in overestimation of intervention effects. Blinding in dry needling trials, particularly blinding of participants and therapists, is a practical challenge; therefore, specific effects of dry needling have yet to be determined. Despite this, dry needling is widely used by health practitioners internationally for the treatment of pain. This review presents the first empirical account of the influence of blinding on intervention effect estimates in dry needling trials. The aim of this systematic review was to determine whether participant beliefs about group allocation relative to actual allocation (blinding effectiveness), and/or adequacy of blinding procedures, moderated pain outcomes in dry needling trials. Methods: Twelve databases (MEDLINE, EMBASE, AMED, Scopus, CINAHL, PEDro, The Cochrane Library, Trove, ProQuest, trial registries) were searched from inception to February 2016. Trials that compared active dry needling with a sham that simulated dry needling were included. Two independent reviewers performed screening, data extraction, and critical appraisal. Available blinding effectiveness data were converted to a blinding index, a quantitative measurement of blinding, and meta-regression was used to investigate the influence of the blinding index on pain. Adequacy of blinding procedures was based on critical appraisal, and subgroup meta-analyses were used to investigate the influence of blinding adequacy on pain. Metaanalytical techniques used inverse-variance random-effects models. Results: The search identified 4894 individual publications with 24 eligible for inclusion in the quantitative syntheses. In 19 trials risk of methodological bias was high or unclear. Five trials were adequately blinded, and blinding was assessed and sufficiently reported to compute the