Although noninferiority trials in cardiology have become more commonplace and offer some advantages vs more traditional superiority trials, clinicians must remain aware of certain pitfalls and heed certain caveats when designing noninferiority experiments and interpreting their results, according to a review published in Heart.
Whereas conventional clinical trials generally aim to demonstrate superiority of a treatment to either placebo or standard therapies, noninferiority studies focus on providing evidence that a new therapy is not inferior to existing treatments, and often use active comparators vs placebos. Noninferiority trials require a different approach than do superiority trials, with alterations in their hypotheses, design, and interpretation. Importantly, noninferiority trials are all carried out under the premise that the new treatment, in certain crucial aspects, offers distinct advantages over standard therapies.
The new treatment in question is considered “not worse” or “similar” to the standard treatment, and therefore tolerable or acceptable for use if the noninferior margin, a preset margin of difference between the conventional and new therapies, is not exceeded. This margin can be defined using either absolute difference or relative difference methodologies, each of which carries unique advantages and disadvantages. The standard therapy event rate must be assumed when using the former approach, although this is not necessary when employing the latter method. Sequential testing is often performed for both superiority and noninferiority, although doing so carries its own challenges.
One instance in which noninferiority trials are frequently necessary is when ethics prevent further testing of a new therapy against a placebo. However, certain assumptions inherent to such trials may actually be incorrect and untested, including the beliefs that the new therapy is superior to placebo and that it offers advantages compared with standard treatments. And as mentioned earlier, despite its utility (particularly when testing for noninferiority is followed by superiority testing) and initial demonstrations of noninferiority, sequential noninferiority testing can often produce progressively less effective treatments.
In addition, noninferior results are favored by poor-quality studies as well as intention-to-treat analyses; this presents a hurdle to investigators, as such intention-to-treat analyses are the preferred approach for interpretation of randomized trials. As a consequence, both intention-to-treat and per protocol analyses are recommended when evaluating noninferiority trials, and conclusions from both types of analysis must agree for noninferiority to be declared. Finally, the focus of interpretation in noninferiority trials should be on effect sizes and confidence limits, rather than the more traditionally used P values.
Common noninferiority trial deficiencies include inappropriate use of this type of design and false claims of equivalence or noninferiority compared with nonsignificant superiority trial results. Researchers also frequently fail to offer proper rationales for noninferior designs or margins.
“Clinicians should be aware of the pitfalls of non-inferiority trials and not accept non-inferiority on face value,” cautioned the authors.
Leung JT, Barnes SL, Lo ST, Leung DY. Non-inferiority trials in cardiology: what clinicians need to know [published online October 31, 2019]. Heart. doi:10.1136/heartjnl-2019-315772