Keeping people in a study unaware of who got which treatment.
Blinding is the practice of concealing group allocation from one or more parties in a trial — participants, treating clinicians, outcome assessors, or data analysts — so that knowledge of who received the intervention cannot shape behaviour, care, or reporting. A study is described as single-, double- or triple-blind depending on how many of these groups are kept unaware, though the terms are used inconsistently across the literature so it's more useful to check exactly who was blinded.
Without blinding, participants who know they got the active treatment may report better outcomes through expectation or demand effects, clinicians may treat groups differently, and assessors may score outcomes more favourably for the group they believe is receiving the 'real' intervention. This is especially important for subjective outcomes like pain or function, where performance and detection bias can inflate apparent effect sizes; it matters far less for hard, objective endpoints like mortality.
Papers should state explicitly who was blinded (participants, therapists, assessors, analysts) and how blinding was achieved and maintained — for example sham devices, identical-looking interventions, or a separate assessor uninvolved in delivering treatment. Many trials also report a blinding check or credibility rating, and risk-of-bias tools such as the Cochrane RoB 2 assess blinding of outcome assessors as a distinct domain from blinding of participants and personnel.
In exercise, manual therapy and rehabilitation trials, true blinding of participants and treating clinicians is frequently impossible since people can usually tell whether they are exercising, being manipulated, or receiving a placebo-like sham. Reviewers should be wary of studies that claim 'double-blinding' without specifying who was actually unaware, and should instead check whether outcome assessors and data analysts were blinded even when participant blinding could not be achieved.
This guide was auto-drafted and is pending editorial review.