Saturday, September 06, 2014

Full Analysis Set and Intention-to-Treat Population in Non-randomized Clinical Trials?

Intention to treatment principle has now been an routine term in the statistical analysis for randomized, controled clinical trials. If a publication is for a randomized, controled clinical trial, it is almost universal that the intention to treatement principle will be mentioned even though the actual analysis may not exactly follow the intention to treat principle in some studies.

Strictly speaking, the intention to treatment principle indicates that the intention to treatment population 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. See one of my early articles and the presentation on ITT versus mITT.

According to ICH E9 “STATISTICAL PRINCIPLES FOR CLINICAL TRIALS”, Full Analysis Set (FAS) is identical to the Intentio-to-Treat (ITT) population. It states:

“The intention-to-treat (see Glossary) principle implies that the primary analysis should include all randomised subjects. Compliance with this principle would necessitate complete follow-up of all randomised subjects for study outcomes. In practice this ideal may be difficult to achieve, for reasons to be described. In this document the term 'full analysis set' is used to describe the analysis set which is as complete as possible and as close as possible to the intention-to-treat ideal of including all randomised subjects.

Here both FAS and ITT population are tied to the randomization. However, in the real world, there are also a lot of non-randomized trials, for example, a clinical study without concurrent control, an early phase dose escalation study without a concurrent control, a long-term safety follow up study where all subjects receive the experimental medication. In these situations, since there is no randomization, it is inappropriate to define an ITT population even though the general principle should remain the same, ie, to preserve as many subjects as possible to avoid bias. The issue is that without randomization, what will be trigger point for defining the ITT population? It looks like that the trigger point could be the time of the administration of the first study medication. Instead of allocating subjects in ITT once randomized’, the subject is in ITT ‘once dosed’. This seems to be the case in the following example, according to CSL’s RIASTAP summary basis of approval, the ITT population was defined for a study without concurrent control and without randomization. It implied that the ITT population includes all subjects who received the study medication, which is essentially the same as the Safety population.

For non-randomized studies, it may be better to use Full Analysis Set instead of ITT population. It seems to be logical to define the full analysis set to include any subjects who receive any amount of the study medication. If this definition is used with the trigger point being the first dose of the study medication, most likely, the full anlaysis set and the safety population will be identical. It is not uncommon that we define two populations that is identidical, but use it for different analyses. For safety analyses, the safety population is used; for efficacy analyses, full analysis set is used.

Another term we can use in non-randomized studies is Evaluable Population which is usually defined as any subjects who receive any amount of the study medication and have at least one post-baseline efficacy measurement. Evaluable population in non-randomized clinical trials is similar to the modified ITT population in randomized clinical trials where some randomized subjects are excluded from the analysis with justifiable rationales.

While the ICH E9 did not use the term ‘modified Intention-to-Treat’, the following paragraphs are intented to provide the guidelines or examples when the subjects can be excluded from the full analysis data set or Intention to treatment population:

“There are a limited number of circumstances that might lead to excluding randomised subjects from the full analysis set including the failure to satisfy major entry criteria (eligibility violations), the failure to take at least one dose of trial medication and the lack of any data post randomisation. Such exclusions should always be justified.Subjects who fail to satisfy an entry criterion may be excluded from the analysis without the possibility of introducing bias only under the following circumstances:(i) the entry criterion was measured prior to randomisation;
(ii) the detection of the relevant eligibility violations can be made completely objectively;
(iii) all subjects receive equal scrutiny for eligibility violations; (This may be difficult to ensure in an open-label study, or even in a double-blind study if the data are unblinded prior to this scrutiny, emphasising the importance of the blind review.)
(iv) all detected violations of the particular entry criterion are excluded.
In some situations, it may be reasonable to eliminate from the set of all randomised subjects any subject who took no trial medication. The intention-to-treat principle would be preserved despite the exclusion of these patients provided, for example, that the decision of whether or not to begin treatment could not be influenced byknowledge of the assigned treatment. In other situations it may be necessary to eliminate from the set of all randomised subjects any subject without data post randomisation. No analysis is complete unless the potential biases arising from these specific exclusions, or any others, are addressed.
In some situations, it may be reasonable to eliminate from the set of all randomised subjects any subject who took no trial medication. The intention-to-treat principle would be preserved despite the exclusion of these patients provided, for example, that the decision of whether or not to begin treatment could not be influenced byknowledge of the assigned treatment. In other situations it may be necessary to eliminate from the set of all randomised subjects any subject without data post randomisation. No analysis is complete unless the potential biases arising from these specific exclusions, or any others, are addressed.
Because of the unpredictability of some problems, it may sometimes be preferable to defer detailed consideration of the manner of dealing with irregularities until the blind review of the data at the end of the trial, and, if so, this should be stated in the protocol.”
In summary, while the general principle is the same, the different terms may be preferred to be used depending on a study being a randomized or non-randomized.

Randomized studies
Non-randomized studies
ITT
Full Analysis Set
Safety
Safety
mITT
Evaluable


Interesting talks about the Intention to treatment principle:


3 comments:

Anonymous said...

Thank you for this post. How does PP (per-protocol) fit in all of this ?

Web blog from Dr. Deng said...

if you collect the protocol deviations during the study, you can still define a per-protocol population as a subset of full analysis set. You can then use the per-protocol population to perform sensitivity analyses.

Unknown said...

All your articles are quite informative and "TO THE POINT".. Thank you for sharing !!