ANIMAL AND HUMAN TESTING IN SKIN CARE 209 the 71 given ointment Z, who showed exacerbations of their lesions while using the ointment. The final challenge patch test, given two weeks after the use test was completed, confirmed the sensitization potential of oint- ment Z, the organic-sulfur-containing product. Thus, to combine it with a use test is in our opinion the best way to ex- tend the range of the patch test. We have recently used another effective approach. A very well known and respected ointment product had its formula modified to enhance permeability. Presumably, the new formula was as bland as the old, and conventional patch tests were reported to have shown this. In use, the material had caused trouble, particularly when applied to irritated skin. Apparently some chemical in the preparation was more irritating to the true skin than it was to the epidermis. To test this premise, we intentionally scarified the skin, then ran simultaneous patch tests on the intact and injured skin. The data obtained from 50 individuals of the test panel are shown in Table 3. The behavior difference TABLE 3--MODIFICATION OF TEST SITE (50 INDIVIDUALS) ---Old Ointment• ,----New Ointment--• Test Retest Test Retest Intact Skin: Erythema 0 5 5 12 Erythema and edema 0 0 0 0 Scarified Skin: Erythema 1 1 26 16 Erythema and edema 0 1 17 26 is striking. The new formula material was apparently both irritating and sensitizing to the abraded skin. This explained the trouble caused in use. Having acquired the data, at this point we should discuss the statistical validity of the results of patch tests of this type. Actually even with the recommended panel of 200 individuals if they were patch tested and showed no reaction it would only mean that there would still be likelihood that there might be quite a high incidence of reactions in the general population, i.e., as high as 1.5 per 100 persons. Obviously, if one were to rely upon statistics, pretesting would be rather impractical. It has been determined mathematically just how many individuals should be tested in a given panel and these conclusions are shown in Fig. 5 (8). Here the maxi- mum anticipated number of positive patch test reactions in the general population has been calculated in terms of negative, 1 plus and 2 plus posi- tive patch tests in a sample population of given size. The samples range from 30,000 to 10 individuals and as is seen, the anticipated number of re- actions from a completely negative patch test series, range from 0.01 per cent to 26.9 per cent respectively. The answer here is that one never at- tempts to interpret patch test data purely statistically. Irritation and

210 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS sensitization studies are comparisons which must always be controlled by simultaneous trial of a similar agent that has been in use over many years. Another major objection that has been raised to the use of experimenta- tion on human skin, is the individual variation in skin response. For ex- ample, it has been shown that those individuals with eczema or the individ- uals who have a tendency to acquire eczema apparently become sensitized much more readily than do those with normal skin. The instance of sensi- tization is almost double in the diseased skin. Data on this point, orig- inally presented by Sulzberger (9) are shown in Fig. 6. Referring back to Statistical Calculatlon• Maximum Anticipated Reactions in General Public from analysis of Prophetic Patch Test Data Test Population (sample size) 30,000 Individuals 10,000 5,000 1,000 200 100 50 20 l0 Number of Positive Reacti? ? •_n Test None One Two_ 0.01% , o.o3% o .06% o.3o% 1.5% •.o % 5.a % 13.9 % 26.9 % o.o54 o .o•o o.45% 2.2 % 4.4 o• 8.8 % 20.6 • I 0.06% , i 2.9 % 5.s % 27.0 Modified from Henderson, C,R, and Riley, E. C. Figure 5.---Statistical interpretation patch test to general population. our own work on the combined use and patch test, it will be recalled that we too have shown that the diseased skin becomes sensitized much more readily than the normal (1). Another factor that cannot be discounted is that an ingredient which may show no signs of being an irritant or sensi- tizer when presented in one vehicle becomes quite a noxious agent when presented in another. Most frequently this difference in behavior can be attributed to damage of the intact skin due to either solvent irritation or direct penetration. The difference in irritability of the epidermis and the dermis was clearly shown by our work on the abrasion patch test. Manu- facturers often neglect this influence of the vehicle. Simple change of an ointment base from petrotatum to one of the newer polyethylene glycol-