CONSORT 2010 and Controversies Regarding Assessment of Blindness in RCTs

Abstract

Blinding is one of the key design features of randomized clinical trials (RCTs) . Studies not involving blinding could yield biased esti- mates of the effect of treatment. A meta-analysis found that double- blind RCTs found 14% lower treatment effect on average than simi- lar RCTs not described as double-blind. During past years many re- searchers have emphasized the need for more research in the area of blinding. For example, sub-item 11b of the 2007 CONSORT statement suggested the need for an assessment of blinding. We have been waiting for a revision and extension of this item, which would encourage people to collect and report more data and share lessons learned. Yet, disappointingly, it was eliminated from CON- SORT 2010. As stated by the CONSORT authors, the rationale for this elimination was interpretational and measurement difficulties. Since the claim of internal validity for RCTs with a control arm rests on the assumption of appropriate blinding, we cannot afford to lose the CONSORT's in- centive for improvement and change. It seems contrary to an evi- dence-based approach to avoid obtaining data because we have to struggle with interpretation and measurement, which is a common and natural problem in various scientific fields. The philosophy of let us give up because it is difficult to do or interp- ret may not be well justified in professional scientific communities, especially when some reasonable statistical methods and clinical guidelines are available or can be developed.