What Is Blinding? Keeping People From Knowing Which Treatment They Got

Knowing the assignment influences results through several routes that operate independently of the treatment. Participants who believe they received an active treatment report improvement, which is the placebo response and is a genuine measurable effect on reported symptoms and on some physiological measures. Participants who believe they received nothing may report less improvement or leave the study. Those administering the treatment may behave differently, offering more attention or adjusting care. Those assessing outcomes may judge ambiguous results in the direction they expect, which is documented and substantial for subjective measures. And analysts making choices during the analysis can be influenced by knowing which group is which. Blinding addresses each of these by concealing the assignment from the people concerned, and the more of them who are blinded the more of these routes are closed.

Terminology varies between fields and the substance does not:

• •Single blind, where participants do not know their assignment
• •Double blind, where the people administering the treatment do not know either
• •Triple blind, extending to the outcome assessors and sometimes to the analysts
• •Open label, where everyone knows, which is sometimes unavoidable and always a limitation
• •Blinded outcome assessment in otherwise open trials, which is a partial measure worth having
• •Reporting how well the blinding held, by asking participants to guess their assignment, which is good practice and is rarely done

Many important questions cannot be studied this way and the workarounds are imperfect. Surgery cannot easily be concealed, though sham procedures have been used in a small number of trials and have produced striking results, with several established operations shown to perform no better than a sham incision, which is among the strongest arguments for doing it where ethics permit. Behavioural and psychological interventions cannot be hidden from the person receiving them, so those trials blind assessors instead. Treatments with distinctive side effects break their own blinding, since participants experiencing a characteristic effect correctly infer their group, and this is a documented problem for drugs with noticeable effects. Device trials, physiotherapy, diet and exercise all face versions of the difficulty. The honest response is to blind what can be blinded, to measure whether it held, and to report the limitation.

A distinct safeguard is frequently confused with blinding and is arguably more important. Allocation concealment means that whoever enrols a participant cannot know or predict which group the next person will be assigned to, which prevents the enrolment decision being influenced by that knowledge. Without it, a clinician who can see the next allocation may steer a sicker patient towards the treatment they believe in, or delay enrolling someone until the allocation suits, and either destroys the comparability that randomisation was meant to produce. Studies comparing trials with and without adequate concealment find that inadequately concealed trials report substantially larger treatment effects on average, which is direct evidence that the problem is real. Concealment applies at enrolment and blinding applies afterwards, so a trial can have one without the other, and reports should state both separately.

Blinding a drug trial requires a control that is indistinguishable from the treatment, and producing one is harder than it appears. The placebo must match in appearance, taste, smell and route of administration, and matching an injection, an inhaler or a treatment with a distinctive taste requires deliberate work. What the placebo contains matters and is frequently unreported, since an inactive substance may not be inert and several cases exist where the chosen filler had effects of its own. Active placebos, which reproduce side effects without the therapeutic action, address the unblinding problem and are rarely used because they complicate ethics and supply. And the placebo response itself is a real phenomenon with its own literature, varying by condition, by how the treatment is presented and by what the participant expects, which means the comparison group is not an absence of treatment but a different treatment.

Concealment closes several routes at once, covering participant expectation, differences in care, biased assessment of ambiguous outcomes and choices during analysis. Each additional group blinded closes another route. Treatments with distinctive side effects break their own blinding, and sham surgery trials have shown several established operations to be no better than an incision.