The therapeutic Delta-equivalence. The large sample case
Résumé
Several approaches are given in the literature for therapeutic equivalence (the two one-sided test approach, the power approach, the confidence and Westlake intervals approaches, the UMP test approach) yielding distinct definitions. This paper is aimed at passing beyond this heterogeneity in order to propose a standard definition of therapeutic equivalence. We firstly show that there can be no absolute definition, and that we have to define $\Dvalence, where $\Dis a thresholdrdrd that has to be chosen from clinical considerations. Then we study the links and inconsistenci- es of the different approaches in the case of large samples (which allows us to come back to calculations with normal distributions). From the obtained results, we are able to propose the two one-sided test definition as a standard definition. We replace the p-value computation by that of an a posteriori limit value of $\Ds standard -i- definition is simple, it is identical to the UMP test definition in case of one-sided equivalence, and it is close to it in case of two-sided equivalence. We finally propose a strategy for equivalence trials and we initiate the discussion on the medical areas where equivalence should be investigated.
Domaines
Autre [cs.OH]
Loading...