REGRESSION TECHNIQUES TO SOLVE UNEQUAL CHALLENGE POTENCY PROBLEMS IN ANTITUBERCULOSIS VACCINE ASSAYS

Abstract
Direct comparisons between experimental antituberculosis vaccines, not tested in the same mouse assays, are hazardous, for challenge "virulence" and host resistance cannot be ideally controlled between experiments. Standard linear regression techniques, adapted to take full account of such wanderings in challenge infectivity and natural resistance of the host, are described. Their use permits valid comparisons between vaccines not assayed in common. The criteria are noted by which conclusions may be drawn regarding relative potency of two vaccines tested in separate sets of experiments.

This publication has 1 reference indexed in Scilit: