Firstname: Ludwig A.
Institution: University of Hannover
Street: Herrenhauser Str.2
Authors: Hothorn, L.A.
Title: Demonstration interlaboratory similarity of dose-response assay s by modification of interaction contrasts
Abstract: Interlaboratory studies are widespread used pre-clinics, e.g. for toxicological in-vitro assay (1). Common statistical approaches for interlaboratory studies (2) are inappropriate for such dose-response designs, because not the reproducibility of a single replicated measurement is of interest but the similarity of dose-response effects. This can be tested by an interaction test for the two-way layout DOSE (Control, D1,..., Dk)* CENTER. The interaction F-test in the parametric set-up is inappropriate because of its global nature, but all-pairs tetrad interaction contrasts (3) estimate the needed local information. However, only many-to-one comparisons are interesting for the factor DOSE. Moreover, incremental trend contrasts, or isotonic contrasts assuming a monotonous dose-response relationship will be discussed in terms of interaction contrasts. Using liver weight data form a real interlaboratory immunotoxicological study, the several multiple interaction contrast types will be compared. For all approaches SAS programs are made available.
References: 1) Gee P, Sommers CH, Melick AS, Gidrol XM, Todd MD, Burris RB, Nelson ME, Klemm RC, Zeiger E. (1998). Comparison of responses of base-specific Salmonella tester strains with the traditional strains for identifying mutagens: the results of a validation study. Mutat Res. 412,115-30.
2) Horwitz W, Britton P, Chirtel SJ. (1998) A simple method for evaluating data from an interlaboratory study. J AOAC Int 81,1257-65
3) Gabriel, K. R. , Putter, J. , and Wax, Y. (1973), Simultaneous confidence intervals for product-type interaction contrasts, JRSS B35 , 234-244