Blinding

Blinding means keeping people involved in a trial from knowing which treatment each participant received. It protects the study after randomisation, in the same way allocation concealment protects it during randomisation. Five groups can be blinded: participants, the clinicians delivering treatment, the data collectors, the outcome assessors, and the statisticians who analyse the results. Schulz and Grimes argued in 2002 that the labels single, double and triple blind are used so inconsistently that trials should simply state who was kept unaware. Blinding matters because knowledge changes behaviour. An unblinded participant reports symptoms differently, an unblinded clinician adds co-interventions, and an unblinded assessor reads borderline findings in the expected direction. The size of that effect is measurable. Across 250 trials analysed in 1995, those without double blinding exaggerated odds ratios by about 17%, and a 2008 study spanning 146 meta-analyses found roughly 25% exaggeration for subjective outcomes, with little bias for hard outcomes such as death. CONSORT requires trials to report who was blinded and how. Some interventions cannot be blinded at all, including exercise, fasting and most surgery. In longevity and nutrition research that is the honest limit: a capsule can be blinded, an eating pattern cannot, and the evidence should be weighted accordingly.

Last reviewed: August 25, 2026

This definition is educational and is not medical advice, a diagnosis, or treatment. Talk to a doctor about any health decisions. Read our full medical disclaimer