J. Soc. Cosmetic Chemists, 16, 145-153(1965) Some Safety Statistical Aspects of Cosmetics of the LINCOLN J. GERENDE, Ph.D.* Presented before the Third Congress of the I.F.S.C.C., June 21-26, 106•, New York City Synopsis--The complaint rates resulting from the use of cosmetics, i.e., the number of com- plaints per number of units sold, is not a statistically valid method of assessing the safety of products. On the other hand, a properly selected random sample involving the population at risk can be used to predict the incidence of irritation resulting from a product. This can be achieved through a customary test or a paired comparison procedure. It appears most practical to setup a maximum tolerable irritation rate before a product is marketed. Tests of the type described can then be used to determine whether the product under consideration warrants marketing. Cosmetic preparations are undoubtedly formulated in the light of available experience with regard to the problems of skin irritation and hypersensitizing substances by taking every precaution to eliminate all possible sources of trouble. Unfortunately, that is not an absolute guarantee against complaints from the consumer. It is doubtful if such a guarantee can ever be given. One may assume that in nature identical conditions produce identical results. Practically speaking, however, the postulated uniformity is not reflected in reality. What is observed is statistical regularity. Hence, it is no surprise if a product turns up an occasional complaint, even though very large numbers of persons use the product without adverse consequences. When a product is formulated containing a new ingredient, or a mix- ture of new ingredients, the chemist is in the position of seeking informa- * Dept. of Preventive Medicine, University of Pittsburgh, Pittsburgh, Pa. 145

146 JOURNAL OF THE SOCIETY OF COSMETIC CHEMISTS tion regarding its potential to irritate or sensitize the skin of the con- sumer. Moreover, it is helpful to have some quantitative estimate of the proportion of such responses in the population of consumers. If one desires information about some characteristic of a qualitative or quantitative nature which exists in a population, it is necessary first to know what population is involved. In the application of the theory of statistics to living beings, the term "population" should be qualified frequently by speaking of the "population at risk." This means nothing more than the group of individuals among whom the event of interest is most likely to occur. Thus, to get the birth rate, say, the population is defined as females between the ages of 15 and 45. The obvious reasons for so limiting the population is that the likelihood of pregnancy among females under 15 or over 45 years of age is relatively small. The birth rate then indicates the number of babies born per female at risk of giving birth to a child. Similarly, in the case of a new rinse for blonde hair, it is desired to determine the irritation potential among possible users. Is it likely that men will be among the consumers? Or, Negroes? It would be ex- pected that Caucasian women who are natural or artificial blondes would buy this product. Moreover, there would be some age restrictions on the population at risk since adults are the only users of such hair prepara- tions. On the other hand, if the present trend continues, it is likely that there will not be many grey-haired women among the elderly so they are possible customers. Defining a population at risk is a matter of setting up a rule or a number of rules which serve as a screening device. THE FALLACY OF COMPLAINT RATES In the cosmetic industry, irritation complaint rates have been formu- lated on the basis of the ratio of complaints to units of product sold. Such a rate may have some economic value, but it provides no informa- tion about the incidence rate of irritation among consumers. The reason for this obviously is that the product is not at risk of experiendng the event. People form the population at risk of suffering an adverse reac- tion to the product. Moreover, although ordinarily a unit of the product is not used up at a single application, it is nevertheless true that for many cosmetic products a consumer may purchase several units in the course of a year. To illustrate the difference this makes, consider the following example: three million units of a given product are sold in a year, and at the end of the year 120 complaints are made to the pro- ducer. Then the irritation complaint rate is 120/3,000,000 or 4/100,000.