Abstract:
|
Equivalence of univariate endpoints are common in the clinical industry. However, it is important to consider multivariate equivalence since most pharmacokinetic (PK) parameters are highly correlated. Least squares procedures like the union intersection method are available, but, not implemented due to lack of clear understanding of these procedures. Additionally, most PK analysis are on phase 1b/2 trials with very small samples. The effect of outliers can exaggerate (or diminish) the true association. A robust analogue of the least squares procedure is presented for sensitivity analysis.
|