Seriously Scientific Logo
LIVE

Science Puzzle

Who Knew What in the Trial?

Scientific Thinking Charge ⚡⚡
Who is allowed to know who got the real drug? OPEN no blinding patient KNOWS doctor KNOWS both can bias the result SINGLE BLIND patient does NOT know doctor KNOWS doctor can still bias it DOUBLE BLIND patient does NOT know doctor does NOT know sealed code held by a third party The patients cannot know, or the placebo effect skews it. But why must the DOCTOR not know either?
Fig. 1: Blinding the patient is obvious. Blinding the assessor is the step people forget.

In a drug trial, patients are not told whether they received the real drug or a placebo. This is sensible: a patient who knows they got the real thing may report feeling better simply because they expect to.

But in a double-blind trial, the doctors assessing the patients are also kept in the dark. Why is that necessary?

The Answer

Because the person doing the measuring is part of the instrument, and an unblinded assessor is a biased instrument. This holds even when the doctor is entirely honest and actively trying to be fair.

Consider what an assessor actually does. Many clinical judgements are not a number read off a machine; they are calls. Is that rash "moderate" or "mild"? Did the patient's walking improve? Is this side effect worth recording? A doctor who knows this patient received the promising new drug will, without any intention to deceive, tend to resolve those borderline calls in the drug's favour. This is observer bias, and it is unconscious by definition. Asking people to try harder to be objective does not remove it.

It also changes behaviour, not just judgement. An unblinded doctor may spend longer with the treatment group, encourage them more warmly, or follow them up more diligently. Any of that can produce a real improvement that has nothing to do with the drug.

This is why the code is held by an independent third party and only broken once all outcomes have been recorded. The design does not rely on anyone being virtuous. It removes the information that makes bias possible in the first place, which is a far more robust strategy than trusting people to overcome a bias they cannot detect in themselves.

Where blinding is genuinely impossible, as in most surgical or exercise trials, the usual solution is to blind the assessor even though the participant obviously cannot be, which recovers most of the benefit.

The principle: Double blinding. The assessor must not know the group assignments, because observer bias is unconscious and alters both judgement and treatment even when the assessor is entirely honest.