PRESERVATIVE EFFICACY TESTING 3 Challenge test through linear regression (5,6), pharmacopeial (1-3) and CTFA (4) meth- ods. Twenty-gram portions of the samples were transferred to sterile tubes and individu- ally inoculated with 0.1 ml of the standardized microbial suspensions. The inoculated tubes were stirred, and each test preparation was submitted to determi- nation of the number of viable microorganisms by the pour-plate method. The dilutions were made using a solution with 0.85% (w/v) of sodium chloride and 0.4% (w/v) of calcium chloride. The medium was tryptic soy agar (Difco Laboratories) with 0.5% (w/v) of lecithin and 0.07% of polysorbate 80. The same procedure was performed with 20 ml of the saline solutions. The number of surviving microorganisms was determined after 2, 4, 24, and 48 hours. The D-values were then calculated with a coefficient of correlation whose linear regres- sion was higher than 0.9000. A similar test was simultaneously performed whose sur- viving microorganisms, however, were determined after 7, 14, 21, and 28 days. Evaluation of the preservative systems neutralization. Lecithin 0.07% (w/v), polysorbate 80 0.5% (w/v), and calcium chloride 0.4% (w/v) were added to the diluent (0.85% solution of sodium chloride) in order to neutralize the preservative systems. The efficacy of the neutralization of the preservative systems was tested by separately inoculating the test microorganisms (1 ml of standardized suspensions containing 100 CFU/ml) into the product diluted to 10 -•. The surviving microorganisms were subsequently determined through the pour-plate method, which led to satisfactory results when the obtained CFU was similar to that of the microbial suspension. RESULTS The results obtained proved the efficacy of the neutralizing agents used. The D-values obtained for different microorganisms are shown in Table I, and the linear regressions Table I The D-Values Obtained by the Linear Regression Method Applied to One Cosmetic Formulation Without Preservative and Eight Others With Different Preservative Systems D-values (hours) Shewanella Bzzrkholderia Escherichia Staphylococczzs Bacillzzs Formula pzztrefaciens cepacia coli azzrezzs sp. I --* .... 2 4.8 18.6 -- 11.6 -- 3 17.9 -- 28.9 12.1 -- 4 -- 37.6 11.5 6.3 -- 5 11.2 -- 16.9 6.0 -- 6 5.6 -- 4.1 4.8 -- 7 5.0 58.8 14.6 8.1 -- 8 1.0 5.2 5.6 4.8 -- 9 0.9 1.3 7.5 1.7 -- * The D-value was not calculated (coefficient of correlation was lower than 0.9000).

4 JOURNAL OF COSMETIC SCIENCE of the same tests are shown in Figure 1. The number of surviving microorganisms after 48 hours and after 7, 14, 21, and 28 days are presented in Table II. DISCUSSION The evaluation of preservative efficacy consists of a challenge test performed against some standard microorganisms and some other ones isolated from the products or production environment. S. aureus was included in this work due to the application of the cosmetic product on the skin B. cepacia was utilized since it can metabolize many kinds of raw materials and is resistant to antimicrobials (13,14) S. putrefaciens and Bacillus sp. are part of the natural bioburden. All the formulations studied, with or without preservative systems, were tested by the linear regression method, which permitted the assessment of their D-values. All the microorganisms employed, except S. aureus, were considered nonpathogenic, as the tested product was for skin use. The data obtained from the tests whose formulas had no preservative system resulted in inadequate performance, since the reduction of the viable organisms was only observed in Bacillus sp. Among the 40 assays performed with the eight formulas inoculated with S. putrefaciens, B. cepacia, E. co/i, S. aureus, and Bacillus sp., the D-value couldn't be determined in 13 of them, since the coefficients of correlation were lower than 0.9000 (Table I), the minimum value acceptable. In spite of this, the plot of the logarithmic number of the surviving organisms per gram, in regard to time, provided information about the antimicrobial activity of the included preservative system (Figure 1). The results obtained from assays with Bacillus didn't provide a D-value in any sample. However, some reductions of viable organisms were obtained in all the samples after 48 hours, and all the preservative systems were then tested in agreement with the requirements for sporeformer microorganisms, revealing their bacteriostatic effect. Even in the formula without any preservative some reduction was observed. When analyzing the D-values obtained by comparison of the requirements (D-value -- 4 hours for pathogenic bacteria and D-value -- 28 hours for nonpathogenic bacteria, yeast, and mold) only formula 9 was in agreement. The other samples that failed the test were formulas 3, 4, 5, 6, and 7 against B. cepacia 2 and 3 against E. coli 4 against S. putrefaciens and 2, 3, 4, 5, 6, 7, and 8 against S. aureus. Formula 9 can be considered the most effective among the eight studied formulas. Formula 8 was also effective, except for S. aureus. Although the D-value for Bacillus sp. hasn't been calculated, the rate of viable microorganism reduction at different times is higher in formula 8 than in formula 9 (Figure 1). When compared to the European Pharmacopeia (1) criteria, the data obtained from formulas 2, 6, 8, and 9 were adequate. According to the CTFA (4) specifications, formulas 2, 4, 5, 6, 7, 8, and 9 (Table II) were also in agreement. Formulas 2, 7, 8, and 9 showed similar behavior under the same circumstances, which makes it impossible to select the best preservative system among them. It is also difficult to determine the most efficient system between samples 8 and 9 if one analyzes the results according to the European Pharmacopoeia (1) specification. The possibility of foreseeing the time for self-sterilization of any bioburden in the product (6) is another advantage of the linear regression method. Since the D-value must be --4 hours when the pathogenic microorganism is tested, the time required for