By ClinicalStudyConnect.com Research Desk | Last updated September 2026
What Does “Intention-to-Treat” Mean in a Clinical Trial?
Intention-to-treat (ITT) analysis is a method where researchers count every participant’s results in the group they were originally assigned to, even if that person stopped treatment, switched groups, or never took the treatment at all. The alternative — analyzing only the people who actually completed treatment as planned — is called per-protocol analysis. When a trial report states that results were analyzed “by intention to treat,” it means the comparison between groups reflects everyone who was randomized, not just the people who stuck with the plan.
Why Researchers Keep People in Their Original Groups
Randomization is what allows a trial to fairly compare a treatment to a placebo or another treatment. When participants are randomly assigned, the two groups start out similar in age, health status, and other factors that could affect the outcome. If researchers later move people out of their assigned group — for example, dropping anyone who stopped taking the study drug — that balance can break down. People who stop treatment early often differ in meaningful ways from those who continue, such as experiencing side effects or seeing no benefit. Removing them selectively can make a treatment look more effective or safer than it actually is.
Keeping everyone in their original group preserves the value of randomization and gives a more realistic picture of how a treatment performs in practice, including cases where people did not follow the protocol exactly.
Terms to Know
- Intention-to-treat (ITT) analysis: Every randomized participant’s outcome is analyzed in their original group, regardless of whether they completed the assigned treatment.
- Per-protocol analysis: Only participants who completed the study as originally planned are included, which can exclude people who dropped out, switched treatments, or deviated from instructions.
- Modified intention-to-treat (mITT): A variation of ITT that excludes a narrowly defined group, such as participants who never received a single dose, while still keeping everyone else in their assigned group. Trial reports should specify exactly how “modified” was defined.
- Dropout or attrition: Participants who leave a study before it ends, whether due to side effects, lack of perceived benefit, relocation, or other reasons unrelated to the study.
- Crossover: When a participant assigned to one group ends up receiving the other group’s treatment, whether by protocol design or by choice.
Intention-to-Treat vs. Per-Protocol: What Each One Can and Cannot Show You
Neither approach is automatically “correct” — they answer different questions, and well-reported trials often present both.
- Intention-to-treat answers: “What happens if people are offered this treatment under real-world conditions, including the fact that some won’t finish it?” This tends to be the more conservative estimate, since it keeps the groups as randomization created them and limits the risk of bias from selective dropout.
- Per-protocol analysis answers: “What happens when someone takes the treatment exactly as intended?” This can be useful supplementary information, but on its own it risks overestimating benefit or underestimating harm, since it excludes people who left the study for reasons that may be connected to the treatment itself.
- A pattern worth watching for: if a trial report emphasizes per-protocol results while downplaying or omitting an ITT analysis, that is a reason to look more closely at how many participants were excluded and why.
A Field Guide: Finding the Analysis Population in a Trial Report
- Look in the Methods section, under “Statistical Analysis” or “Outcomes.” Well-reported randomized trials should explicitly state whether the primary analysis was intention-to-treat, modified intention-to-treat, or per-protocol. Reporting guidance from groups like the CONSORT-SPIRIT initiative exists specifically to standardize how trials disclose this kind of methods detail.
- Check the participant flow diagram. Most randomized trial reports include a flowchart showing how many people were randomized, how many completed the study, and how many were excluded from analysis and why. This is often the clearest place to see the gap between “randomized” and “analyzed.”
- Note how dropouts were handled statistically. ITT analyses still need a way to account for missing outcome data from people who left early — common methods include carrying forward a participant’s last known result or using statistical modeling to estimate likely outcomes. The report should describe which method was used.
- Compare group sizes at randomization versus at analysis. A large difference between the number of people randomized and the number included in the final analysis is worth noting, regardless of which analysis method was used.
- See if both ITT and per-protocol results are reported. When available, comparing the two can show whether the findings are consistent regardless of analysis method, which strengthens confidence in the result — or whether they diverge, which raises questions worth exploring further.
Disclosure: This page contains affiliate links. If you buy through them, we may earn a commission at no extra cost to you.
What This Does Not Tell You
Knowing that a trial used intention-to-treat analysis does not by itself tell you whether the trial was well-designed, adequately sized, or free of other sources of bias. It also does not tell you why participants dropped out — that context matters and is usually found in the discussion or limitations section of the report. A trial can use ITT analysis correctly and still have significant limitations elsewhere.
Medical Disclaimer
This article is for educational purposes only and explains a research methodology concept. It is not medical advice and should not be used to evaluate whether a specific treatment is appropriate for you. Do not start, stop, or change any treatment based on this information. Speak with a qualified healthcare provider about your individual situation, and contact local emergency services for any medical emergency.
Primary source: CONSORT-SPIRIT Statement (current successor to consort-statement.org). Supporting source pending — NBK305514 was assigned but does not address intention-to-treat methodology and has been removed rather than misapplied; needs replacement before publication. Related reading on Clinical Study Connect: Randomized Controlled Trials Explained and Study Populations, Dosing, and Outcomes.