Intention-to-Treat Analysis: Why Dropouts Still Count

ITT keeps every randomized participant in the originally assigned group for the primary analysis ².
That approach preserves baseline balance and reflects the real-world consequences of offering treatment ³.
ITT does not guarantee unbiased results when missing data are handled poorly ⁴.

The missing capsules still belong in the result

A supplement trial can look simple on the label: 200 people were assigned to magnesium, 200 to placebo, then sleep was measured eight weeks later. The catch is that not everyone takes the capsules. Some forget. Some stop because of stomach upset. Some buy a different magnesium product at the store. The tempting move is to analyze only the people who did everything correctly. That feels fair, but it can quietly break the main protection that random assignment gave the study.

Random assignment means people are placed into groups by chance, so the groups start out balanced in known and unknown ways. Intention-to-treat analysis protects that balance by keeping participants in the group they were originally assigned to, even if they missed doses, switched, stopped, or crossed over. In plain terms: once the trial assigns you to a group, your outcome stays with that group for the main analysis.

The surprise: ITT may include people who never took the treatment

That sounds wrong until you name the question. ITT is not asking, “What is the biological effect among perfect users?” It is asking, “What is the effect of assigning this plan to real people?” That is often the better question for policy, clinical advice, and everyday decisions, because plans fail in predictable ways. A supplement that works only when every dose is taken, causes many people to stop, and has no benefit once missed doses are counted may not be as useful as a clean-looking analysis suggests.

This is why major trial reporting standards care about whether authors show how many people were randomized, how many were lost, and how many were included in the analysis. CONSORT, the reporting standard for randomized trials, specifically focuses on transparent reporting of participant flow and analysis groups because readers need to see whether the claimed analysis really kept people where randomization put them.

What ITT does and does not solve

ITT helps prevent a common source of bias: removing inconvenient participants after the trial starts. But it does not magically fix missing outcome data. If someone leaves the trial and no final result is measured, researchers still need a plan for handling that missing information. The International Council for Harmonisation E9(R1) addendum connects ITT to a broader idea called an estimand, which means the exact treatment question the trial is designed to answer. In that framework, the ITT idea often matches a “treatment policy” question: what happens when people are followed and analyzed regardless of whether they fully followed the planned treatment.

The one decision to make today: when you read a randomized supplement study, find the participant flow diagram or methods section and check whether the main result says participants were analyzed in the groups they were assigned to. If the paper only reports “completers,” treat the result as a best-case picture, not the main real-world answer.

When you'll see this

The term in the wild

You are reading a trial of melatonin for sleep and see that 18 percent of participants stopped taking capsules before the final visit.

If the main analysis is ITT, those participants should still count in the groups they were assigned to, as long as their outcome data can be included or handled using a prespecified method.

This keeps the result closer to what happens when people actually try melatonin, including people who stop because it does not help or causes grogginess.
A paper says “per-protocol population” in the abstract and reports only people who took at least 80 percent of capsules.

That is not the same as ITT. It answers what happened among more adherent participants, after excluding people who did not follow the plan closely.

The result may be useful, but it should not replace the main randomized comparison when you are judging whether the intervention works as an assigned plan.
In a CONSORT flow diagram, 500 participants were randomized, but the primary analysis includes 421.

The missing 79 people matter. The paper should explain who was excluded, why, and whether the analysis still matches the ITT principle.

A small line in the flow diagram can change how much confidence you place in the headline result.
A clinician reviews a weight-loss supplement trial where many people in the active group stopped because of nausea.

An ITT analysis counts those people in the active group rather than removing them because they did not finish.

That makes tolerability part of the practical result, which is exactly what a real user needs to know.

Myths vs reality

What people get wrong

Myth

ITT means everyone took the treatment exactly as planned.

Why people believe it The name sounds as if it tracks intention or motivation, when it actually tracks the treatment group assigned by randomization.

Reality

ITT can include people who missed doses, stopped early, switched treatments, or never started. That is the point: the analysis follows assignment, not perfect behavior.
Myth

If a paper says “intention-to-treat,” it definitely analyzed every randomized participant correctly.

Why people believe it The specific named cause is reporting shorthand. “ITT” became a credibility signal in abstracts, while the actual participant flow and exclusions may be buried in the methods section.

Reality

Some papers use the label loosely. CONSORT 2010 reported that trial reports sometimes claimed ITT while excluding participants or failing to analyze them as allocated.
Myth

ITT always gives the best estimate of the treatment’s pure biological effect.

Why people believe it Readers often want one number to mean “does it work,” but trials can ask different legitimate questions.

Reality

ITT usually estimates the effect of assigning or recommending a treatment plan in real conditions. A per-protocol analysis may better address perfect-use biology, but it gives up some protection against bias.

For athletes, ITT is especially important when supplement studies involve side effects, taste, stomach tolerance, or complicated timing. A product that looks effective only after excluding people who could not stick with it may be a poor choice during training or competition, where adherence problems are part of the real decision.

Frequently asked

Common questions

When should I prefer an ITT result over a per-protocol result?

Use the ITT result when deciding whether a treatment plan is useful in real life. Use the per-protocol result as a secondary clue about what may happen among people who can follow the plan closely.

Can ITT be used outside drug trials?

Yes. ITT applies to randomized studies of diets, exercise programs, supplements, apps, and behavioral interventions whenever people are assigned to groups and some do not follow the plan.

What does “modified ITT” mean?

Modified ITT usually means the authors excluded some randomized participants before analysis, often because they never received a dose or lacked any follow-up data. The exact rule matters because “modified” is not used consistently across studies.

Does ITT make a study more conservative?

Often it makes benefits look smaller than a completer-only analysis, because nonadherence and stopping are counted. But it is not automatically conservative in every trial, especially when missing data are handled poorly.

What should I look for after seeing the phrase ITT in a paper?

Look for the randomized number and the analyzed number in each group. If those numbers differ, the paper should clearly explain who was missing and how their outcomes were handled.

Sources

Intention-to-Treat (ITT) Analysis in brief

What Intention-to-Treat (ITT) Analysis means

The missing capsules still belong in the result

The surprise: ITT may include people who never took the treatment

What ITT does and does not solve

Putting Intention-to-Treat (ITT) Analysis to work

Sources