N-of-1 trial
- Study design
A trial in a single patient: two treatments alternate several times, in random order, to find out which works better for that person.
What it is
An N-of-1 trial is a repeated crossover trial in one person: two treatments (or treatment and placebo) alternate over several pairs of periods, with the order in each pair randomised and, when possible, blinded. It answers an individual question: which option helps this patient more. It only works in chronic, stable conditions, with treatments whose effect starts quickly and fades soon after stopping.
How it is built
- Patient and clinician agree on the question and the outcome that matters (for example, pain from 0 to 10 points).
- Three or more pairs of periods are planned, with washout if needed.
- In each pair the order of A and B is drawn at random.
- The patient records the outcome in each period.
- A and B periods are compared within each pair and overall.
Worked example
- A patient with stable knee osteoarthritis compares two analgesics, A and B, over 3 pairs of 2-week periods in random order.
- Mean pain (0 to 10 points): pair 1 A 4, B 6; pair 2 A 5, B 5; pair 3 A 3, B 6.
- Differences B − A: 2, 0 and 3 points; mean = (2 + 0 + 3) / 3 = 1.67, rounded to 1.7 points in favour of A.
- A was better in 2 of the 3 pairs and worse in none.
- Reading: for this patient, A lowers pain about 1.7 points more than B; with the clinician, they decide to continue A.
N-of-1 trial versus other ways of deciding
| Aspect | N-of-1 trial | Classic crossover trial | Usual trial and error |
|---|---|---|---|
| Patients | One | Several, one crossover each | One |
| Repetitions | Several A/B pairs | One A/B pair | None planned |
| Random order | Yes, in each pair | Yes | No |
| Question | What helps this patient? | What helps on average? | Clinical impression |
| Bias | Low if blinded | Low | Expectation and regression to the mean |
Markers
- N-of-1
- Paired periods
- Washout
When to use it
- Chronic pain, insomnia, stable asthma, symptoms of rare diseases with no large trials.
- When group evidence does not fit the patient well, or a long-term treatment is in doubt.
- Useful for deciding, on evidence, to stop a treatment that may no longer help.
Common pitfalls
- Using it in conditions that change fast or resolve.
- Long-acting treatments without enough washout: carry-over mixes the periods.
- Running a single pair of periods: the effect cannot be told apart from chance or natural variation.
Test yourself
Think of the answer, then tap the question to see it.
What must a condition be like for an N-of-1 trial?
Chronic and stable, with a treatment whose effect starts quickly and fades soon after stopping.
Why are several pairs of periods repeated?
To tell the treatment effect apart from natural variation and chance.
To whom does the result of an N-of-1 trial apply?
To the patient studied; only by pooling many N-of-1 trials can an average effect be estimated.
Connections
Educational content on epidemiological and statistical methods; the examples use made-up numbers to teach the calculation. Not a clinical or legal guide: current Colombian rules are in the official sources.
Sources: Chapter 23: Including variants on randomized trials (Cochrane, 2024); Chapter 8: Assessing risk of bias in a randomized trial (Cochrane, 2024).
Spotted a mistake?We are always improving for the community. Tell us what is wrong and our team will review it.Report a mistake
Representative model, for educational purposes only.