Design non-inferiority

Both of these elements are causing nervousness amongst regulators. So much so that, for example, the FDA Anti-Infective Drugs Advisory Committee (AIDAC) have recently recommended that the non-inferiority design should no longer... 

A finding of no significant difference in outcome between two treatments is ambiguous unless a non-inferiority design is used. Two-arm non-inferiority designs require much larger sample sizes to test adequately the statistical hypothesis of no true difference in efficacy between a test and reference drug. 

The results of a non-inferiority design trial can lead to one of the following conclusions ... 

In a paper commemorating the 10-year aimiversary of ICH ElO, Rockhold and Enas (2011) discussed the relevance of ElO for safety non-inferiority trials and noted that, despite the inherent complexities in their design, it is likely that the use of non-inferiority designs for safety will continue. Safety non-inferiority trials were also discussed in a review paper by Fleming (2008). 

Is a difference sought or is equivalence the objective International conference of harmonisation (ICH E9 ) makes it clear that it is vital that the protocol of a trial, designed to demonstrate equivalence or non-inferiority contains a clear statement that this is its explicit intention (ICH E9, Section 3.3.2). In the past if a trial failed to show that a new treatment gave benefit compared to a standard, it was commonplace to claim that the new treatment was therefore as effective as the standard. Such an argument is no longer acceptable as will be discussed in Section 8.5.6. 

Non-inferiority trials are becoming more and more common as time goes on. This in part is due to the constraints imposed by the revised Helsinki Declaration (2004) and the increasing concern in some circles regarding the ethics of placebo use. These trials however require very careful design and conduct and we will discuss this whole area in a subsequent chapter. 

One of the most difficult aspects of the design of equivalence and non-inferiority trials, with the exception of bioequivalence, is the choice of the margin(s). 

Moving in the opposite direction, that is concluding non-inferiority in a superiority trial, is much more difficult, as this would generally require pre-specification of a non-inferiority margin. Such pre-specification would usually not have been considered in a trial designed to demonstrate superiority. However, if a conclusion of non-inferiority would be a useful outcome then it could be appropriate to consider such pre-specification. Switching from superiority to non-inferiority, however, presents further problems. Assay sensitivity may well be one reason why... 

Non-inferiority trials are more difficult to design assay sensitivity and the choice of non-inferiority margin are just two of the issues that would additionally need to be considered. 

Designing the trial as a non-inferiority evaluation may give a negative perception within and outside of the clinical trial team. 

The correct design, analysis and interpretation of non-inferiority trials -remember conventional p-values have no role... 

An adequately powered trial initially designed to demonstrate the noninferiority of a new compound compared with a standard treatment can also demonstrate the superiority of the new compound provided that a statistically significant difference in favor of the new1 treatment was detected. However, trials intended to show the superiority of one of the treatments cannot be interpreted, in the case of a statistically non-significant difference, as showing non-inferiority of the new1 chemical entity compared with the standard treatment. 

The profile of the Lagrange multiplier along the non-inferior solution curve is used to obtain the optimal solution when a unit cost of exergy is specified. Using the two-objective analysis we can obtain a much better understanding of the process design under the present uncertain conditions with respect to energy. 

If only active comparator trials were available, then the trial objectives would have to be very precise from the beginning of a trials programme as to whether superiority, equivalence or non-inferiority is being tested.Demonstration of non-inferiority in turn depends on designing a trial with sufficient sensitivity, as it has to rely on indirect evidence that a trial is capable of showing a difference, without prior availability of placebo-controlled data (see Section 6.3.2.12)... 

Chen, M.-H., Ibrahim, J. G., Lam, R, Yu, A., and Zhang, Y. (2011) Bayesian design of non-inferiority trials for medical devices using historical data, Biometrics 67,1163-1170. 

Non-Inferiority Study Design and Analysis for Safety Endpoints... 

Note the use of the word "important" as a modifier of "difference" in the previous paragraph. It is generally not feasible, or even possible, to demonstrate conclusively that there is no difference at all between two treatments. Therefore, non-inferiority and equivalence are often defined through the use of a margin, typically designated by 5, to indicate the smallest difference between treatments that would be considered "important." (In the cases of equivalence, different margins for superiority and inferiority are sometimes used.)... 

In a sense, virtually, all clinical trials can be considered to be safety noninferiority trials, since virtually, all trials collect safety information and therefore provide evidence as to whether or not there is an important difference in safety between treatments. However, there are two key distinguishing characteristics of the trials that we refer to as safety non-inferiority trials in this chapter a safety endpoint is designated as primary and a noninferiority margin is prespecified. In the next section, we will discuss some of the issues associated with non-inferiority trials in general, focusing on the factors that are unique to safety non-inferiority trials. 

While preservation of effect is an important consideration for efficacy noninferiority trials and is discussed in both the FDA and EMA guidance documents, it is not relevant to the design of a safety non-inferiority trial. 

When the control rate is known, these three approaches will give consistent conclusions regarding non-inferiority. The issue arises when the control rate differs from the hypothesized rate. Suppose now that the trial was designed under the assumption that the response rate in the control group is 0.2 but in reality, in the context of the non-inferiority trial, the true response rate is 0.1. In this case, the three metrics can result in inconsistent conclusions. Consider the response rate in the experimental group that would be at the boundary of the non-inferiority definition using the three margins described previously. 

To our knowledge, there is no consensus on the most appropriate metric for binary or time-to-event endpoints, and, therefore, this is a factor that must be carefully considered by an investigator designing a non-inferiority trial with such endpoints. 

Non-inferiority trials are designed to demonstrate that one treatment has an effect that is either superior to that of another treatment, or similar to it. Safety non-inferiority trials are non-inferiority trials with a primary safety endpoint and with a prespecified margin. Non-inferiority trials for efficacy and safety endpoints have some similarities, but also some fundamental... 

---