1. CQ Deng, PhD PPD Development Research Triangle Park, NC 27560
Intention-to-Treat and modified Intention-to-Treat Analyses in Clinical Trials
CQ Deng, PhD
PPD Development
Research Triangle Park, NC 27560

2. Confusion about Intention-to-Treat?
Statistical analyses are based on intention-to-treat basis
ITT population includes all randomized patients
…… who take at least one dose of study drug
…… who take at least one dose of study drug and have at least one post-baseline efficacy measurement
…… who complete three treatment cycles
ITT population includes all patients who take at least one dose of study medication   

3. Confusion about Intention to Treat?  
“…The division definition of (modified) intent-to-treat is all patients who are randomized to treatment, and have the infection confirmed by microscopy and culture.
The sponsor’s definition of intent-to-treat further limits this to cases with a clinical signs and symptoms score greater than two and requires at least one non-missing post-baseline efficacy assessment…”
FDA CDER statistical review of NDA

4. What is the Intention-to-Treat analysis?
Includes all randomized patients in the groups to which they were randomly assigned, regardless of their adherence with the entry criteria, regardless of the treatment they actually received, and regardless of subsequent withdrawal from treatment or deviation from the protocol
Fisher, LD et al. Intention to treat in clinical trials in Statistical Issues in Drug Research and Development. Edited by Peace KE (1990)

5. Intention-to-treat analysis
Full analysis set. Analyze once randomized!
Analyze as randomized, not as treated
Preserve the initial randomization, keep the baseline comparability among treatment groups
Minimize bias. Prevent the conscious or unconscious attempts to influence the results of the study by excluding the patients
Conservative for estimates of the treatment difference
Preferred for trials to show a difference between two treatments
Ignore noncompliance, protocol deviations, withdrawal, and anything that happens after randomization

6. Reasons for patients to be excluded from ITT:
Further tests after randomization show the patient is ineligible or misdiagnosed
The patient does not receive any of their allocated treatment
The patient takes the wrong study drug
The patient receives some, but not all their allocated treatment
The patient is not assessed for the outcome of interest, such as response

7. Practical definition of ITT
Includes all patients:
who were randomized
who were known to take at least one dose of treatment
who provided any follow-up data for one or more key efficacy variables
based on the randomized treatment, not the treatment actually received  (if mis-randomization rate is less than 5%)
Gillings and Koch (1990) The application of the principle of intention-to-treat
to the analysis of clinical trials. Drug Information Journal

8. Overuse/Misuse of ITT In non-randomized trial For safety evaluation  
In non-randomized trial
No randomization, No ITT!
For safety evaluation
ITT analysis may underestimate the incidence rate
In some bioavailability/bioequivalence trials
No reason to include the subjects who did not take study drug
Follow “as treated”, not “as randomized”

9. Overuse/misuse of ITT In equivalence/non-inferiority trials  
In equivalence/non-inferiority trials
As intention-to-treat analyses tend to dilute an effect between treatment, an ITT analysis in an equivalence trial may make the treatments appear to be more similar than they actually are.
For equivalence trials a ‘per protocol analysis’ could be regarded as ‘conservative’ and therefore is often given as much emphasis as an ITT.

10. Overuse/misuse of ITT Assuming Safety population = ITT population  
Assuming Safety population = ITT population
“ITT population includes all randomized patients who take at least one dose of study drug”
“Safety population includes all randomized patients who take at least one dose of study drug”
ITT = Safety population ?
In the case of randomization error:
NSafety Total = NITT Total
NSafety for each treatment group  NITT for each treatment group
Assuming Per-protocol is purely a subset of ITT

11. What is the mITT?
Modified intention-to-treat (mITT), may also be called quasi ITT, is a subset of the ITT population and allows the exclusion of some randomized subjects in a justified way.   

12. Some examples of mITT definitions
“The mITT population included all patients who were randomized and had a positive culture of (pathogen) at baseline.”
“The mITT population included all randomized patients who developed (symptoms) and took at least one dose of study medication.”
“The mITT population included any patient who was randomized, took at least one dose of study medication during stabilization period 2 (SP2), maintained a stable dose of 2400 mg/day during SP2, had baseline migraine headache data, and at least 1 day of migraine headache evaluations during SP2.”*
* Mathew NT et al (2001) Efficacy of Gabapentin in Migraine Prophylaxis. Headache   

13. modified Intention-to-Treat
Not a full analysis set - a subset of ITT
Analyze as randomized, not as treated
Many practical ITT definitions may be called mITT
Popular in antimicrobial / anti-infective trials
Multiple mITT populations can be defined for a single study: Clinical mITT, Microbiological mITT

14. Situation when mITT is appropriate  
When the disease diagnosis is not immediately available at randomization or at start of treatment
Patient
developing
symptoms
Randomization
Initiate the
treatment
Suspected
patient
for the trial
confirmatory diagnosis results are available
For the acute disease caused by a bacteria, virus, or fungus, the confirmatory
diagnosis usually takes several days. The randomization and the start of treatment
cannot wait until the confirmatory diagnosis results are available. SARS (Corona virus),
Bird Flu (H5N1 subtype of influenza A), Pneumonia…

15. Situation when mITT is appropriate  
When patients initiate the treatment
Patient develops
symptoms
Takes study med
Randomization
Dispense drug
Patient
has history
of disease
Patient never
develops symptoms
No treatment
initiated
ITT population includes all randomized patients
mITT population includes all patients who developed symptoms and took
at least one dose of study medication
Example: Recurrent genital herpes, cold sores (recurrent herpes labialis) trials

16. Situation when mITT is appropriate
When the period between randomization and start of the medication is long . The longer the period, the more likely the patient will change his/her mind   
Drug shipped
to the site
Randomization
Patient decides not
to participate in the trial
Patient has no knowledge of which treatment group he/she is in.
Whether or not patient takes the study medication is random.
There is no merit to say there is any potential bias by excluding those patients
who did not take the study medication.

17. Caveat when using mITT
Exclusion of the patients should be in a justified way, not at will
Patient exclusion is not associated with patient characteristics
or clinical outcome
ITT (full analysis set) is suggested for sensitivity analysis
Other sensitivity analyses may also be employed   

18. Summary
ITT
Full analysis set, include all patients who were randomized
As randomized, not as treated
Primary analysis population in superiority trials
ITT can be overused or misused
mITT
A subset of ITT, not a full analysis set
Preserve some ITT features
In some situations, mITT is more appropriate
When using mITT, ITT is suggested for sensitivity analysis
Popular in antimicrobial / anti-infective trials