The equivalence or non-inferiority problem in systematic reviews

Article type
Authors
Lange S, Biester K
Abstract
Background: The proof of a comparable efficacy or harm of 2 or more treatments is a well known problem in randomized clinical trials (RCTs). It has recieved widespread attention in the past. Such trials are called 'equivalence trials' (ETs) or 'non-inferiority trials' (N-ITs). An essential issue in planning ETs or N-ITs is the pre-specification of the 'irrelevant difference', which quantifies the amount of difference or inferiority which is to be considered tolerable. Comparability can then be assumed, if a difference larger than this quantity can be excluded by a confidence interval for the treatment difference or an appropriate statistical test. It is obvious that the equivalence or non-inferiority problem also arises in systematic reviews (SRs). Astonishingly, thus far, this issue seems to be neglected in the methodological literature with regard to SRs.

Objectives: It was the aim of our study to get an impression whether the equivalence or non-inferiority problem is an issue in SRs.

Methods: A random sample of 100 SRs was drawn from the Cochrane Library, 25 each for the periods 1998-1999, 2000-2001, 2002-2003, and 2004-2005. To be included, the reviews had to deal with therapeutic interventions, and the reviews had to summarize the results of at least 5 single trials. Protocols and /or updates were excluded. Two independent reviewers examined (i) whether equivalence or non-inferiority was stated as a problem of the reviews themselves, and (ii) whether irrelevant differences were pre-specified.

Results: (i) Equivalence or non-inferiority was stated as a problem of the review itself in only 4 reviews. (ii) An irrelevant difference was pre-defined in one review.

Conclusions: In SRs, equivalence or non-inferiority is only rarely mentioned as a problem of the reviews themselves. While we found several statements in regard to which differences between treatments were regarded as 'comparable', an irrelevant difference was pre-defined in only one review, as it is a part of good statistical practice in RCTs, when it is the aim to show comparability. Methodological advice is urgently necessary when considering how to deal with this issue in SRs.