DETECTION THRESHOLDS OF CAPSAICIN 159 caused too-significant sensations on two subjects with self-declared very sensitive skin. Thus, a second study was carried out on 11 of the 21 subjects with two more diluted solutions (3 .16 x 10- 5 % and 1 x 10-4%). In the latter case, the test was painless, even for the two aforementioned subjects with very sensitive skin. Moreover, the second study allowed us to assess repeatability. On the 11 test subjects, detection thresholds never differed more than one level between two test sessions carried out at a two-week interval. STATISTICAL METHODS First, descriptive statistics and graphics were carried out. Percentages of positive re- sponse for all items of the questionnaire were tabulated. The distribution of detected capsaicin concentration was described using simple bar charts, as the frequency of self-declared sensitive skin and the mean age in subpopulations ranked by detected capsaicin concentration. Then, different inferential statistics were calculated depending on the nature of the data. For each item of the questionnaire, the association with the detected capsaicin concentration was tested using Kendall's 'T 6 statistic. The relationship between self-declared sensitive skin and the detection threshold was tested using the Cochran-Armitage trend test. In addition, the influence of age on detected capsaicin concentration was tested using the non-parametric Jonckheere-Terpstra trend test. All results are reported with two-sided p-values. The significance level u was set to 5%. Statistical analysis was carried out using SPSS software (Version 11.00 SPSS Inc., Chicago, IL). RESULTS DETECTION THRESHOLDS OF CAPSAICIN During the initial stages (cleansing or double application of the vehicle), no subjects reported discomfort due to the vehicle, and thus no premature withdrawals occurred before the first capsaicin application. Consequently, the capsaicin detection threshold was determined in the total study population (n = 150). Figure 2a shows the distribution of the study population according to capsaicin detection thresholds. Note that the population is distributed among six possible levels: the five capsaicin solutions and the level "none." This distribution shows clearly that there were several different detection thresholds of capsaicin in this general adult female population. As illustrated, the range of detected capsaicin concentration was at least of two log units because 27 .3% (N = 41) of the panelists did not reach their detection threshold. More noteworthy about the distribution was its shape. Considering the previous works on sensitive skin typology, one could have expected a unimodal distribution with high frequencies for the mean concentrations and low values for the extreme concentrations. As shown, it was clearly not the case. The pattern of the obtained distribution suggests the existence of at least a bimodal distribution. RELATIONSHIP WITH SELF-DECLARED SENSITIVE SKIN The association with detection threshold was established for all the items of the ques- tionnaire. Table II displays Kendall's 'T 6 values and their significance. Note that the

160 JOURNAL OF COSMETIC SCIENCE 30% n = 41 25% c� 20% a- .. t' j C 15% •- ID .b :I ar 10% IL 5% 0% 3,16.10" 1 % a Perceived capsai In concentration 100% "D 80% iii' "'i'� 60% �:ii: : fD - m 0 t':e C ! m m 40% :I UJ 20% IL 0% 3,16.10 "' % None b Perceived capsaicin concentration Figure 2. Capsaicin detection thresholds. (a) Distribution of the capsaicin detection thresholds in terms of incidence. (6) Relationship between the capsaicin detection threshold and self-dedared sensitive skin. This figure shows the incidence of self-reported sensitive skin in the six subpopulations classified according to detected capsaicin solution. relationship with the detection threshold was significant for 13 of the 20 items. Inter- estingly, the most statistically significant link concerned the item "sensitive facial skin" (Kendall's 'T 6 p 0.001). Detection thresholds were globally more strongly associated with self-declared skin reactivity to environment (cold, wind, fast changes in tempera- ture) than to cosmetics. The association was highly significant with skin reactivity to