Intention-to-treat analysis
Intention-to-treat (ITT) analysis keeps every participant in the group they were randomised to, whatever happened afterwards, which preserves randomisation (allocating people to groups by chance).
Remember
Once randomised, always analysed: staying in the original group protects the balance that randomisation created and shows how a treatment performs in everyday practice.
Once a trial is over, there are two competing principles for analysing the results. In an intention-to-treat (ITT) analysis, every participant is analysed in the group they were randomised to, regardless of adherence (taking the treatment as instructed). People who switched to the other treatment or dropped out stay in their original groups.
Randomisation (allocating participants to groups by chance) gives groups that are similar at the start. People who stop a drug or switch to another are often different from those who keep going, for instance because of side effects. If those people were dropped or moved to another group, the groups can end up differing in important ways.
ITT is also practical: it shows how the treatment performs in everyday care when some patients do not take their tablets. That is why it is called a pragmatic (real-world) analysis. ITT therefore estimates effectiveness, which is how a treatment works in everyday use, rather than efficacy, which is how it works when taken perfectly.
A per-protocol analysis, in contrast, analyses only patients who completed the protocol (the trial plan) as intended. It asks what a drug can do under ideal conditions, but loses the protection of randomisation.
The price is dilution: the measured difference tends to shrink. Stopping treatment and crossover (a switch to the other group's treatment) make the groups look more alike, so ITT tends to be conservative, often giving a difference smaller than the drug's biological effect. Even so, in most licensing trials, those that support approval of a medicine, it is the primary analysis.
One limit remains: ITT is not a cure for missing data, because keeping someone in their original group cannot recover a result that was never measured.
Key facts
- The rule
- Every randomised participant is analysed in their original group, including those who stop, switch or drop out.
- Slogan
- Once randomised, always analysed.
- What it answers
- ITT asks how a treatment performs in routine practice, including stopping and switching.
- Per-protocol analysis
- Only participants who followed the plan are analysed, so it asks what a drug can do under ideal conditions and gives up the protection of randomisation.
- Dilution
- Stopping and switching make the groups end up more alike, so ITT tends to be conservative: it often gives a difference smaller than the drug's biological effect.
Exam traps
- Calling an analysis intention-to-treat does not mean it estimates the effect under ideal conditions, because people who stop or switch treatment still stay in their original groups, so the result reflects everyday practice.
- An analysis that leaves out people who never took a dose is a modified ITT analysis, not the ITT principle itself, because it still excludes people after randomisation.
- ITT is not a way to increase the apparent difference between the groups: some people stop and others switch, so the groups look more alike and the measured difference is usually diluted.
Read the full notes
Tested in 6 SimpleMed+ questions
Practice it in Pharmacology, with the answer and a full explanation after every question.
Practice these questions