N-of-1 Trial

One patient is given repeated, blinded periods of active treatment versus a comparator (or placebo), in a randomised order, with outcomes measured each period. Comparing the patient against themselves removes between-person variation. Suitable only for chronic, stable, reversible conditions; results from several N-of-1 runs can be pooled across patients.

A single patient passing through several randomised, blinded treatment pairs, then a verdict reached by comparing that one person across their own treatment and comparator periods — not against anybody else.

Read it as a within-person experiment. Because the same patient supplies every measurement, fixed differences between people (genetics, comorbidity, baseline severity) cannot distort the result. Randomising the order of the pairs and blinding both patient and assessor guard against time-trends and placebo response.

For the individual in the chair, a well-run N-of-1 sits unusually high on the evidence hierarchy — it can give a more reliable answer for them than a large trial whose “average” patient may not resemble them. It turns a trial-and-error prescribing decision into a small controlled experiment.

• N = 1, but MANY randomised crossover periods
• Each patient is their own control (within-person comparison)
• Needs a chronic, stable, reversible condition + a washout

Where you would (and would not) use it — N-of-1 designs suit chronic, stable, reversible problems such as recurrent neuropathic or chronic pain, stimulant titration in ADHD, or bronchodilator choice in stable asthma, where a patient can be cycled blindly through treatment and comparator and asked which works better for them. It is the WRONG design for an acute, one-off, or irreversible presentation — you cannot re-randomise a STEMI, a major haemorrhage or a cardiac arrest through repeated treatment periods, so N-of-1 has essentially no role in resuscitation-room decisions.

• Carryover between periods without an adequate washout — the previous treatment contaminates the next.
• Treating an N-of-1 result as generalisable — it answers a question about one patient unless many are formally pooled.
• Applying it to an acute or irreversible condition, where repeated crossover is impossible.