The principle of intention-to-treat

Example 7.1 Surgery compared to radiotherapy in operable lung cancer 

Statistical Thinking for Non-Statisticians in Drug Regulation Richard Kay 2007 John Wiley Sons, Ltd ISBN 978-0-470-31971-0 

Option 1 compare the mean survival time of the 90 who received the intended surgery with the mean survival time of the 100 who were assigned to radiotherapy. 

Remember that all 200 patients provide data on the primary endpoint so this analysis ignores the data on ten of those patients those who were assigned to surgery, but did not receive the intended operative procedure. Perhaps we can think of ways of including those data. Options 2 and 3 include these patients. 

---

The principle of intention-to-treat (ITT) tells us to compare the patients according to the treatments to which they were randomised. Randomisation gives us comparable groups, removing patients at the analysis stage destroys the randomisation and introduces bias. Randomisation also underpins the validity of the statistical comparisons. If we depart from the randomisation scheme then the statistical properties of our tests are compromised. 

The previous section clearly indicates the need to conform to the principle of intention-to-treat to ensure that the statistical comparison of the treatment groups remains valid. In practice compliance with this principle is a little more difficult and the regulators, recognising these difficulties, allow a compromise. This involves the definition in particular trials of the full analysis set which gets us as close as we possibly can get to the intention-to-treat ideal. 

The discussion in the previous section regarding the practical application of the principle of intention-to-treat does not, however, give the full picture. While this principle plus consideration of the per-protocol set may clearly define the sets of subjects to be analysed, we still have to decide how to deal with the missing data caused by failure to complete the study entirely in line with the protocol. 

In each of these situations there are clinical arguments that support the use of the particular endpoint concerned. From a statistical point of view, however, each of these endpoints gives an analysis that is subject to bias in a clear violation of the principle of intention-to-treat. We will look at each of the settings in turn. 

What impact are the dropouts and missing data likely to have on the definition of analysis sets and in particular our ability to align with the principle of intention to treat ... 

Conforming to the principle of intention-to-treat to avoid bias - ITT means all randomised subjects or something very close to that... 

Gillings D and Koch G (1991) The application of the principle of intention-to-treat analysis of clinical trials Drug Information Journal, 25, 411-424 Greenwood M (1926) The errors of sampling of the survivorship tables Reports on Public Health and Statistical Subjects, No. 33, Appendix 1. London HMSO Grieve AP (2003) The number needed to treat a useful clinical measure or a case of the Emperor s new clothes Pharmaceutical Statistics, 2, 87-102 Haybittle JL (1971) Repeated assessment of results in clinical trials of cancer treatment British Journal of Radiology, 44, 793-797... 

The term full analysis set was introduced in order to separate the practice of intention-to-treat from the principle, but practitioners still frequently use the term intention-to-treat population when referring to this set. The term modified intention-to-treat population is also in common use within particular companies and also by regulators in some settings where exclusions from strict intention-to-treat are considered. 

The intention-to-treat 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. ... 

The intent to treat principle essentially states that all patients are analyzed according to the treatment they were randomized to, irrespective of the treatment they actually received. Hence, a patient is included in the analysis even if that patient never received the treatment. 

In some of the studies, the statistical evaluation did not strictly follow the intention-to-treat principle. This is deemed to be of minor importance if only a very small number of patients is excluded because of early termination and drop-out rates are balanced across the treatment groups. In a few trials, neither a primary variable was specified nor was the... 

The intent of this chapter is to present a brief review of simple, fundamental physicochemical principles and experimental results which are necessary to understand both the mechanism of adsorption of ionic surfactants from aqueous solutions on oxide surfaces and the action of some simple, fundamental applications. It does not enter into details in the theoretical consideration, nor does it attempt to explain complex industrial uses. Both problems have been thoroughly treated in several review articles and monographs [e.g., 1-10]. Here emphasis is placed on the contribution the adsorption calorimetry makes to the improvement of current understanding of the interactions of ionic surfactants at the mineral-water interface. All experimental data, used for the illustrative purposes throughout this chapter, were obtained at the Laboratoire des Agregats Moleculaire et Materiaux Inorganiques. 

II. I. The main condition offragmentation of molecular systems Assume that our polyatomic system is divided in two regions, QM and MM (Fig. 2) with the intention to perform combined QM/MM treatment of its properties. In some cases it is useful to divide the QM region in two (or more) layers to be treated at a different level of computation (Section III), and/or to introduce a third region around the QM/MM one, the boundary region (BR), in order to take into aecount the interaction with the remaining environment (which is always present, in principle). So far in this paper we... 

The above examples showed that ideal mass action laws require correction in a variety of cases. Ideal mass action laws for defects only apply when there are random distributions and, hence, at very low defect concentrations. A high disorder — consider, for instance, a solution of 5% SrO in lanthanum copper oxide — can certainly no longer be treated as a simple structural perturbation rather for accurate analysis, individual treatment of the mixed phases is necessary. Then, of course, tractable models lose their general character, and the reader s expectations of the correction possibiUties should not be too high. In this section the intention is — with just one exception — to discuss simple concepts, that are valid, when the ideal mass action laws have just lost their validity , that is, we are still limiting ourselves, in principle, to more or less low defect concentrations. In many cases, however, such concepts qualitatively indicate the right direction, beyond the region of strict vahdity. 

It is well known, that even with the best intentions of faithful adherence to a pre-conceived plan, there are instances, where a deviation from the written procedure would be a necessary act, based on a rational decision. If such possibilities were not accounted for in GLP, it could be criticised with good reason that GLP were an inflexible, bureaucratic impediment to truly scientific investigations. It depends, however, on the exact circumstances how such changes are to be treated in a GLP compliant way. The definitions in the OECD Principles distinguish two different cases, namely the Amendment and the Deviation. 

---