IntentiontoTreat

Intention-to-treat (ITT) is a principle for analyzing randomized controlled trials in which participants are kept in the groups to which they were randomized, regardless of adherence, protocol deviations, withdrawal, or crossovers. ITT preserves the balance achieved by randomization and estimates the effect of assigning the intervention under real-world conditions, providing a conservative and generalizable measure of effectiveness.

In practice, ITT requires including all randomized participants in the primary analysis and handling missing outcome data with appropriate methods. Common approaches include multiple imputation or model-based techniques such as mixed-effects models. Simple methods like last observation carried forward are generally discouraged because they rely on untestable assumptions and can bias results. The chosen method should align with assumptions about the missing data mechanism.

ITT is often contrasted with per-protocol or as-treated analyses, which analyze participants according to the treatment actually received. While per-protocol analyses can address questions about the biological efficacy of a treatment, they are more prone to selection bias due to non-random adherence patterns and early withdrawals.

Limitations of ITT include potential dilution of treatment effects when non-adherence or crossovers are substantial, and the dependence of estimates on how missing data are handled. Nevertheless, ITT remains the standard primary analysis in many randomized trials and is commonly required by regulatory agencies and reporting guidelines to provide an unbiased estimate of the effect of assigning an intervention. Additional analyses exploring adherence effects may be reported alongside ITT.