RT Journal
A1 Harris, Alex H.S.
A1 Fernandes-Taylor, Sara
A1 Giori, Nicholas
T1 “Not Statistically Different” Does Not Necessarily Mean “the Same”: The Important but Underappreciated Distinction Between Difference and Equivalence Studies
JF The Journal of Bone & Joint Surgery
JO The Journal of Bone & Joint Surgery
YR 2012
FD March 7
VO 94
IS 5
SP e29 1
OP 4
DO 10.2106/JBJS.K.00568
UL http://dx.doi.org/10.2106/JBJS.K.00568
AB 
Researchers often want to evaluate whether a new medical or surgical treatment is equivalent to an existing treatment. The new treatment may be preferred if its results are equivalent to those of the existing approach in terms of complications or outcomes but it is superior in terms of ease of use, safety, or cost. However, many researchers are unaware that the equivalence of two interventions cannot be established by failing to find a statistical difference between them. This somewhat subtle detail of research design and statistical analysis has very important clinical implications. The goal of this brief paper is to explain the distinctions between the familiar difference trial (or superiority trial) study design and the often more appropriate but much less familiar equivalence study design. These designs have different underlying hypotheses, power calculations, statistical analyses, and conclusions. Claiming the equivalence of two interventions on the basis of a nonsignificant difference in the results of a difference trial and analysis, as is unfortunately common, may lead to incorrect conclusions and inappropriate changes in clinical practice1,2. Erroneous equivalence claims in the medical literature have been reviewed by Greene et al.3.