Impact of blinding on estimated treatment effects in randomised clinical trials: meta-epidemiological study

Dear Editor,

The problem is that this kind of study can't really capture the bias introduced by not blinding. It looks at average effects across studies, where the bias may actually go in different directions in different studies and thus cancelling out. Imagine a poorly blinded homeopathy study. If performed by proponents of homeopathy, the bias will increase the reported effects; if performed by sceptics, the bias will decrease the reported effects. If performance bias is present, non-blinding may cause additional co-interventions in the control group, leading to decreased treatment effects, whereas for subjective outcomes the effect of measurement bias may be opposite.