A Statistical Basis for Using Fewer Rabbits in Dermal Irritation Testing

Abstract

An acceptable and validated in vitro method to evaluate the potential of a chemical to cause dermal irritation does not exist; therefore, in vivo studies remain the only alternative. Currently most laboratories utilize 6 rabbits per test, but this group size may not be necessary to derive the desired information. Data generated from 6-rabbit skin irritation tests of 105 materials were used to determine the ability of irritation scores from all possible combinations of 5-, 4-, 3-, or 2-rabbit subsets to predict the Draize score derived from 6 rabbits. There are 630, 1575, 2108, and 1575 possible combinations of 105 studies for the 5-, 4-, 3-, and 2-rabbit subseta, respectively. We classify materials using a four-level adjectival rating system based on (among other factors) the Draize score, Comparisons indicated that the 5-, 4-, 3-, and 2-rabbit scores were in 96, 94, 91, and 88% agreement, respectively, with the classification assigned on the basis of the 6-rabbit score. The correlation coefficients for randomly selected subsets of 5-, 4-, 3-, and 2-rabbit scores versus the 6-rabbit Draize score were 0.996, 0.994, 0.986, and 0.977, respectively. This study indicated that 3 rabbits per test group allow for adequate assessment of the dermal irritation potential of a chemical. These results also conformed closely to those obtainetd from a previous statistical study using eye irritancy data from 155 chemicals, where 3-rabbit subsets were 94% predictive of the 6-rabbit tests.