Slide 1: The intention to treat principle. How do we use data from participants in clinical trials who don’t follow the protocol?
Slide 2: Crossover. Sometimes, a participant in clinical trial will not receive the treatment to which they are randomized and will “cross over” to another treatment group. Reasons for crossover include: the assigned treatment was inconvenient or hard to tolerate, the patient received another treatment off-label, and a patient was randomized to conservative management but decompensated and needed emergent surgery.
Slide 3: What do we do with the data for patients who cross over? Option one: per protocol: patients are ONLY counted if they followed the protocol (crossovers are removed).
Slide 4: What do we do with the data for patients who cross over? Option two: intention to treat: patients are counted as being in the group they were randomized to, regardless of whether they remained in that arm.
Slide 5: Why is intention to treat so important? Intention to treat preserves prognostic balance between groups. Consider the following example: You’re conducting a RCT of a new antibody, coreimumab, on cardiovascular mortality, with n=500 patients. Let’s suppose coreimumab truly has no effect and is equivalent to placebo.
Slide 6: Example. Patients who drop out are no longer random! They had specific reasons for dropping out that correlate with other lifestyle factors – in general, adherent patients have a “healthy user bias” in that being able to adhere to a trial correlates with other healthful behaviors, such as diet and exercise. Let’s also assume that more people drop out of the treatment group than the placebo group, perhaps because of side effects.
Slide 7: This is called a “consort diagram.” Graphic showing randomization to the coreimumab and placebo groups. Because we already know the treatment has no effect, an equal number of patients should die in each arm.
Slide 8: Question: what happens when more patients are non-adherent in the intervention than control arm?
Slide 9: Graphic showing randomization to coreimumab and placebo and various outcomes based on which patients are adherent, non-adherent, and died. Outcomes shown by per protocol and intention to treat. Per protocol analysis incorrectly concludes that the intervention has a mortality benefit.
Slide 10: Graphic showing randomization to coreimumab and placebo outcomes based on which patients. Adherent vs. non-adherent patients are prognostically different in ways that existed before the study.
Slide 11: Take home.
The main two ways of analyzing non-adherence to trial protocols are per protocol and intention to treat.
Crossover creates prognostic imbalance between groups because the groups are no longer random.
The intention to treat principle states that participants should be analyzed according to the group to which they were randomized, regardless of what treatment they actually received.
Analysis by intention to treat preserves randomization and prognostic balance between groups.
- The Principle of Intention of Treat and Ambiguous Dropouts. Users’ Guides to the Medical Literature: A Manual to Evidence-Based Clinical Practice. JAMA. McGraw-Hill. 2021 May 12. Link.
- Gupta SK. Intention-to-treat concept: A review. Perspect Clin Res. 2011 Jul;2(3):109-12. PMID 21897887.
- McCoy CE. Understanding the Intention-to-treat Principle in Randomized Controlled Trials. West J Emerg Med. 2017 Oct;18(6):1075-1078. PMID 29085540.
- Detry MA, Lewis RJ. The intention-to-treat principle: how to assess the true effect of choosing a medical treatment. JAMA. 2014 Jul 2;312(1):85-6. PMID 25058221.
Tags: intention to treat, per protocol, statistics, stats with core im