Granuloma pouch anti-inflammatory assay: some remarks on calculation and statistical analysis of results.

Abstract

ABSTRACT A statistical analysis is made of the granuloma pouch assay in the rat as a quantitative method for estimating the anti-inflammatory activity of different materials. The problems of calculating the best unbiased mean and estimates of standard error and degrees of freedom on replicate experiments are studied. Use is made of established criteria for deciding whether the means to be combined are, or are not, of the same precision, and the course to follow in each case. Emphasis is laid on the verification of the hypothesis of non-linearity between response and logarithm of dose of materials (steroids) administered. Estimating equations as well as characteristic statistical parameters are presented for all steroids tested. Results of this analysis indicate that there is a linear logdose-response relationship and that the slopes of the estimating equations for all steroids assayed do not differ significantly from that for the standard. The symbol XY50 is defined to signify the calculated dose causing 50 per cent effect in a graded response assay in which the range of response has been by arithmetic conversion arbitrarily established from 0 to 100. The reliability of obtaining estimates of relative potency as ratios of different XY50 is demonstrated theoretically and illustrated experimentally.