Free scoping read. A written assessment of your manuscript in two business days, at no cost.

Start one
Contents

On this page

  1. Preservation of randomisation balance under intention-to-treat analysis
  2. Inflation of apparent treatment effects in per-protocol analysis
  3. Choice of primary analysis by trial type

Academy · Research Writing · ITT vs per-protocol

Intention-to-Treat Versus Per-Protocol Analysis in Randomised Trials: Distinct Analyses for Distinct Questions

Intention-to-treat (ITT) analysis evaluates the trial as conducted, whereas per-protocol analysis evaluates the treatment as intended. Reporting only one of these analyses, or combining them without disclosure, can lead to overestimation of a treatment’s effect. This article explains the distinction and the circumstances in which each analysis is primary.

Pubrica AcademyWritten by a clinical trial methodologist Reviewed bya biostatistician Published10 October 2026 Last checked10 October 2026 Reading time7 minutes Sources1 peer-reviewed

Intention-to-treat (ITT) analysis includes every randomised participant in the group to which they were assigned, regardless of whether they completed or adhered to the assigned treatment. Per-protocol analysis includes only those participants who adhered to the protocol as specified. The two analyses can yield different results, and each addresses a different question.

Learning objectives

  • Why ITT preserves the benefit of randomisation and per-protocol does not.
  • Why per-protocol analysis tends to inflate the apparent effect of an effective treatment.
  • Which analysis is typically primary for superiority trials versus non-inferiority trials.

Common misconceptions

  • “Per-protocol analysis gives the most accurate result, since it includes only participants who received the treatment.” Gupta explains that per-protocol analysis, by excluding non-adherent participants after randomisation, disrupts the balance in known and unknown confounders created by randomisation; participants who withdraw or discontinue treatment often differ systematically from those who continue, so per-protocol results may be biased rather than more accurate.
  • “ITT analysis is always the more conservative choice.” ITT is conservative for demonstrating a treatment effect, because non-adherence dilutes the measured difference towards the null; in a non-inferiority trial, however, this direction is reversed, and dilution towards “no difference” can make an inferior treatment appear non-inferior. For this reason, per-protocol analysis is typically reported as a co-primary analysis in non-inferiority trials.

Preservation of randomisation balance under intention-to-treat analysis

Gupta describes intention-to-treat as analysing participants according to the group to which they were randomised, regardless of the treatment they actually received, protocol deviations, or withdrawal1. This matters because randomisation balances known and unknown confounders between groups at baseline; analysing participants by their assigned group, not their actual exposure, is what preserves that balance in the analysis.

Inflation of apparent treatment effects in per-protocol analysis

Per-protocol analysis restricts the comparison to participants who adhered to the protocol as specified — completing the assigned treatment course without major deviations. Gupta notes that this approach can introduce bias because the reasons for non-adherence or withdrawal are frequently related to the outcome itself (for example, participants experiencing side effects or lack of benefit are more likely to stop), so excluding them after the fact can make the remaining, adherent group appear more favourable than the treatment would perform in routine clinical use1.

Choice of primary analysis by trial type

For superiority trials, ITT is conventionally the primary analysis, precisely because its conservative bias toward the null makes a positive result more credible. For non-inferiority trials, this logic is reversed: because ITT's dilution toward “no difference” can make a genuinely inferior treatment appear non-inferior, regulatory guidance typically expects both ITT and per-protocol analyses to be reported, with consistency between them strengthening confidence in the non-inferiority conclusion1.

Key points

Key points

ITT analysis evaluates participants by assigned group regardless of adherence, preserving the balance achieved by randomisation; per-protocol analysis includes only adherent participants and may be biased towards a favourable result. ITT is conventionally primary in superiority trials, whereas non-inferiority trials typically require both analyses.

Practical application

Practical application

  1. The primary analysis and its rationale should be stated before unblinding. ITT or per-protocol analysis should be pre-specified as primary in the statistical analysis plan, according to the trial type (superiority or non-inferiority).
  2. Both analyses should be reported for non-inferiority trials. ITT and per-protocol results should be presented together, with discussion of whether they agree, rather than only the more favourable result being reported.
  3. The ITT population should be defined precisely. Authors should state explicitly whether the analysis is a strict ITT (all randomised participants) or a modified ITT (for example, excluding participants never dosed), since the two are not interchangeable.

Frequently asked questions

A commonly used variant that excludes randomised participants who never received any dose of the assigned treatment, while still retaining everyone else regardless of adherence — it should be explicitly defined and distinguished from strict ITT.

This is generally discouraged by reporting guidelines including CONSORT, since it omits the protective effect of randomisation and can overstate the treatment's benefit.

It can in some scenarios; however, the more common and better-documented concern is that it overstates benefit by selectively retaining participants for whom the treatment worked sufficiently well for them to continue.

References

  1. Gupta SK. Intention-to-treat concept: A review. Perspectives in Clinical Research. 2011;2(3):109-112. read 10 October 2026. View on PubMed Central

Every source above is a peer-reviewed article, a publisher’s record of one, or a standard issued by the body that owns it. Each carries the date it was last read. Where a standard moves, this article is revised rather than left up with a note.

How to cite this article

Pick your referencing style and copy it. Your university or target journal may vary the punctuation — check its guide if it has one.

Pubrica Academy. (2026, October 10). Intention-to-Treat Versus Per-Protocol Analysis in Randomised Trials: Distinct Analyses for Distinct Questions. Pubrica. https://pubrica.com/academy/concepts-definitions/itt-vs-per-protocol-analysis/

This article is written to be learned from. Cite it if you are referring to what it says; if you are using the evidence behind it, cite the sources listed above instead.

ITT/per-protocol reporting check

For use when drafting the statistical analysis plan.

[ ] Primary analysis (ITT or per-protocol) pre-specified, matched to trial type
[ ] ITT population definition stated precisely (strict vs modified)
[ ] Both analyses reported for non-inferiority trials
[ ] Reasons for any exclusion from ITT stated and justified
[ ] Consistency between ITT and per-protocol results discussed

Pubrica AcademyWritten and maintained by Pubrica’s subject matter experts. Articles are revised when the underlying standards change, and each article states the date on which it was last reviewed.

How Pubrica can help

Pubrica assists in defining and pre-specifying ITT and per-protocol populations appropriate to the trial type before unblinding.

Submit an enquiry Get your analysis populations checked

Explore

Related Academy articles

Selected from the same topic and category, with the most recent first.

Article

The Impact of Electronic Health Records on Research

What electronic health records change about research data, the five quality dimensions reviewers check, and what FDA real-world-data guidance expects…

10 October 2026 · 10 min read

Article

Adverse Drug Effects: Definition and Classification

The precise difference between an adverse event and an adverse drug reaction, the standard mechanistic classification, and why causality has to be…

10 October 2026 · 8 min read

Article

What Post-Marketing Surveillance Is

What post-marketing surveillance is, why pre-approval trials cannot detect every safety signal, the ICH E2E pharmacovigilance planning framework, and…

10 October 2026 · 8 min read

Find something else

Search the Pubrica Academy

Access is free and requires no account. Enter a topic, or open the full library to filter by category, sub-topic or resource type.

Definition of analysis populations before finalisation of the statistical analysis plan

A scoping assessment is free and takes two business days. It provides a written assessment of what the work requires and which specialist would undertake it.

Get a free quote Ask about your own case
Get a free quote

Regulations

Check your project against this rule

  1. 1Your work
  2. 2Your details
  3. 3Send
What do you need?

Material you send is held under signed confidentiality, with access limited to the assigned team and encrypted transfer. It is not used to train anything and is not passed to third-party tools. Ask for the agreement before you send anything you would rather cover first.