Concept Architecture
Concept
Theoretically, Intention-to-Treat Analysis is an analytical principle for randomised controlled trials in which participants are analysed according to the treatment group to which they were originally randomised, irrespective of treatment adherence, protocol deviations, treatment switching or withdrawal. The concept is founded on randomisation theory and causal inference. It exists to preserve the prognostic balance created by randomisation, minimise selection bias and provide an unbiased estimate of treatment effectiveness under real-world conditions.
Mathematically, Intention-to-Treat Analysis is represented within the standard statistical framework of hypothesis testing and regression modelling, with treatment allocation defined by the original randomisation rather than the treatment actually received. Treatment effects are estimated using all randomised participants whenever outcome data are available, and appropriate methods such as multiple imputation or likelihood-based estimation are used to address missing data while preserving the original treatment allocation.
In practice, all eligible randomised participants remain in the primary analysis regardless of treatment compliance, crossover, discontinuation or protocol violations. Statistical analyses compare outcomes according to the assigned treatment groups using appropriate regression, survival analysis or generalised linear modelling techniques. Intention-to-Treat Analysis is regarded as the primary analytical approach for confirmatory randomised clinical trials and provides treatment effect estimates widely used in health technology assessment and health economic evaluation.
Purpose
Used to preserve the benefits of randomisation by analysing participants according to their original treatment allocation, thereby reducing bias and providing an unbiased estimate of treatment effectiveness under routine clinical practice.
Mathematical Formulae
Primary Formula
Treatment effect estimate:
??ITT = ??R ? ??C
where:
- ??ITT = intention-to-treat treatment effect
- ??R = estimated mean outcome for the randomised intervention group
- ??C = estimated mean outcome for the randomised control group
Supporting Formulae
General linear model:
Y = ?? + ??T + �
where:
- T = original randomised treatment assignment
Cox proportional hazards model:
h(t�T) = h?(t) ? exp(?T)
where treatment assignment remains based on the original randomisation.
Related Mathematical Methods
- Randomised Controlled Trial
- General Linear Models
- Logistic Regression
- Cox Proportional Hazards Model
- Multiple Imputation
- Maximum Likelihood Estimation
- Sensitivity Analysis
Example
A randomised trial allocates 500 participants to a new medicine and 500 to standard care. During follow-up, 60 participants assigned to the new medicine discontinue treatment and 25 participants assigned to standard care switch to the new medicine. Under Intention-to-Treat Analysis, all 1,000 participants are analysed according to their original randomised allocation. The estimated treatment effect therefore reflects the effectiveness of the treatment strategy rather than the efficacy among fully adherent participants, providing evidence appropriate for health technology assessment.
Excel Implementation
| Function | Example Formula | Health Economics Application |
|---|---|---|
| IF | =IF(RandomisedGroup="Treatment",1,0) | Classify participants according to original randomisation. |
| AVERAGEIFS | =AVERAGEIFS(OutcomeRange,RandomisedGroupRange,"Treatment") | Calculate mean outcomes for the intervention group regardless of adherence. |
| AVERAGEIFS | =AVERAGEIFS(OutcomeRange,RandomisedGroupRange,"Control") | Calculate mean outcomes for the control group. |
| COUNTIFS | =COUNTIFS(RandomisedGroupRange,"Treatment") | Verify that all randomised participants remain in the analysis. |
| IFERROR | =IFERROR(TreatmentMean-ControlMean,"") | Calculate the intention-to-treat treatment effect. |
VBA (Optional)
VBA can automate intention-to-treat datasets by retaining original treatment allocation, identifying protocol deviations and producing analysis-ready trial summaries.
Sources
- Friedman LM, Furberg CD, DeMets DL, Reboussin DM, Granger CB. Fundamentals of Clinical Trials.
- Gupta SK. Intention-to-Treat Concept: A Review. Perspectives in Clinical Research.
- ICH E9. Statistical Principles for Clinical Trials.
- CONSORT 2010 Statement.
- NICE. Health Technology Evaluation Manual.
- Drummond MF, et al. Methods for the Economic Evaluation of Health Care Programmes.
Related Concepts (2)
Library
Publications
1
Economic Evaluation in Clinical Trials — Glick, Doshi, Sonnad & Polsky, 2nd Edition ed., 2015 (Oxford University Press)
Practical guidance on conducting cost-effectiveness analyses alongside controlled trials, covering trial design, measurement of costs and quality-adjusted life years, handling censored and missing data, and reporting stochastic uncertainty. Volume 4 in the Handbooks in Health Economic Evaluation series.
BookView source →
Frequently Asked Questions (6)
What is intention-to-treat analysis?
A method analysing trial data by original random treatment assignment, regardless of whether participants actually received or completed that treatment.
Source: Friedman LM, Furberg CD, DeMets DL, Reboussin DM, Granger CB. Fundamentals of Clinical Trials. 5th ed. Springer; 2015. doi:10.1007/978-3-319-18539-2.
Why does intention-to-treat analysis keep patients in their assigned group?
Intention-to-treat analysis counts every patient in the group they were randomly assigned to, whether or not they took or completed the treatment. Keeping them there preserves the balance that randomisation created, because dropping patients who deviated, often for reasons linked to their prognosis, would reintroduce the very selection differences randomisation removed. It therefore gives an unbiased comparison of the treatment strategies as they would work in practice, including imperfect adherence. Analysing as randomised protects against bias. Friedman and colleagues (2015) explain this.
Source: Friedman et al. 2015
Why is intention-to-treat analysis used?
Intention-to-treat analysis is used because it preserves the comparability of groups created by randomisation: analysing participants by their assigned treatment, regardless of deviations, prevents the bias that arises if those who do not adhere or complete treatment are excluded or reassigned, since such participants often differ systematically. It also reflects the real-world effect of assigning a treatment, including the influence of non-adherence. By maintaining randomisation and avoiding selection bias, intention-to-treat analysis gives an unbiased estimate of the effect of treatment assignment, making it the standard primary analysis.
Source: Friedman, Furberg & DeMets 2015
How does intention-to-treat analysis differ from per-protocol analysis?
Intention-to-treat analysis includes all randomised participants in their assigned groups regardless of adherence or completion, whereas per-protocol analysis includes only those who adhered to the protocol, excluding non-adherers and those with deviations. Intention-to-treat preserves randomisation and gives an unbiased estimate of the effect of assignment, while per-protocol estimates the effect among those who followed the treatment but risks bias, since adherers may differ from non-adherers. So intention-to-treat is the standard for avoiding bias, and per-protocol provides a complementary, though potentially biased, view of the effect under full adherence.
Source: Friedman, Furberg & DeMets 2015
What are the strengths of intention-to-treat analysis?
The strengths of intention-to-treat analysis are that it preserves the comparability of groups from randomisation, avoiding selection bias from excluding non-adherers or those who deviate; it gives an unbiased estimate of the effect of assigning a treatment; and it reflects real-world effectiveness, where adherence is imperfect. It provides a conservative, policy-relevant estimate of what assigning the treatment achieves in practice. These strengths make intention-to-treat analysis the standard primary analysis for randomised trials, since it protects validity and answers the practical question of the effect of offering a treatment.
Source: Friedman, Furberg & DeMets 2015
What are the limitations of intention-to-treat analysis?
The limitations of intention-to-treat analysis include that, by including non-adherers, it estimates the effect of assigning a treatment rather than the effect of actually receiving it, which can dilute the apparent efficacy, so it may understate the effect of the treatment when taken as intended. It also requires handling missing data for participants who drop out, which involves assumptions. For equivalence and non-inferiority trials, intention-to-treat can be less conservative. These limitations mean intention-to-treat analysis is complemented by per-protocol and other analyses to understand both the assignment effect and the effect under adherence.
Source: Friedman, Furberg & DeMets 2015
Trust Record
Verified by Dr Darrin Baines
British health economist
Professional identity: darrinbaines.org
Verification date: 14 Nov 2025
Content version: 1.0.0
Canonical Identity
- Term code
- HE-ES-CTM-045
Stable URI · Machine-readable · Resolvable · CC BY 4.0