ASSESSMENT OF FACIAL WRINKLES 131 nasolabial groove, and mouth angles) was done using the five-grade standard on photos of the frontal face of 366 females (the first and second groups) by the two specialists. Clinical test of wrinkle-improving agents. A clinical test of an agent for the improvement of wrinkles (Sofina Seraty, Kao Corp.) was carried out in 15 Japanese females. This agent and placebo (a base cream) were topically applied by the subjects to both outer eye corners in the half-face manner daily for about eight weeks. Evaluations of the wrinkles of each subject were done before and after the application period by a specialist, in the double-blind manner, using the nine-grade scale. Statistica/analysis. The influence of age on wrinkling was studied using analysis of linear, logarithmic, and exponential correlation coefficients. Inter-observer agreement and in- tra-observer repeatability were analyzed and quantified by use of the kappa coefficient (11,12), a chance-corrected intraclass correlation coefficient with possible values ranging from -1 (complete disagreement) to +1 (complete agreement). Values above 0.61 are generally interpreted as indicating substantial agreement, values between 0.41 and 0.60 as moderate agreement, and values between 0.21 and 0.40 as fair agreement in general, an agreement level above fair is preferable (13). Table I shows an illustrative agreement matrix (11). In the evaluation of the improvement in wrinkles following the use of a cosmetic agent, differences between the before-use and after-use scores were analyzed by the paired Student t-test. RESULTS COMPARISON OF THE FIVE-GRADE AND NINE-GRADE PHOTOSCALES FOR EYE-CORNER WRINKLES Five general examiners performed a five-grade evaluation and a nine-grade evaluation of wrinkles on photos of the eye corners, according to the same photo scales. More than Table I An Agreement Matrix of Proportions Judge A Score 1 2 3 4 5 Y, fib Judge B 1 fl 1 * fl 2 fl 3 fl 4 fl 5 5',fl b 2 f21 f22 f23 f24 f25 Y, f2b 3 f31 f32 f33 f34 f35 5',f3b 4 f41 f42 f43 f44 f45 Y, f4b 5 f51 f52 f53 f54 f55 Y, f5b Zfai Zfa• Y. fa2 Zfa3 Zfa4 Zfa5 Zf = Nr * Numbers of notations for two-observer agreement. Y, fai: the proportion of units in which the judges agreed. 5',fib: the proportion of units for which agreement is expected by chance. Z(Y, fai x Y, fib) fo-fc f0 = Zfii fc = K - N w NT-f c

132 JOURNAL OF COSMETIC SCIENCE 60% of the five-grade scores of wrinkles at the eye corners in the second group were grade 4 or 5 (Table II), showing relatively many wrinkles. In the nine-grade evaluation, however, subjects with grades 7-9 occupied 38.9% of the third group, and the distri- bution of the scores was relatively even (Table III). The degree of consistency between the five-grade and nine-grade evaluation scores determined by each examiner and the consensus score by calculation of kappa values are shown in Table IV. The kappa values in the five-grade evaluation ranged from 0.400 to 0.657 (mean 0.499), indicating that the agreement was fair to substantial. In the nine-grade evaluation, the kappa value ranged from 0.252 to 0.507 (mean 0.396), indicating fair-to-moderate agreement. Since the distribution of grades and the number of subjects differed between these two evaluations, a direct comparison of the results may not be appropriate. However, the nine-grade evaluation scores were less consistent with the standard scores than were the five-grade evaluation scores. REPRODUCIBILITY OF THE NINE-GRADE SCALE Table V shows reproducibility when the examiner repeated the evaluation of the wrinkles of the same subjects in the third group by the nine-grade photo scale. The kappa values ranged from 0.252 to 0.507 (mean 0.396) in the first scoring and from 0.327 to 0.474 (mean 0.414) in the second scoring, indicating that this nine-grading procedure is reproducible with fair-to-moderate agreement. EFFECTS OF SITE (EYE CORNERS VERSUS FOREHEAD) ON THE WRINKLE SCORES The distribution of the five-grade score for forehead wrinkles is shown in Table VI. In this group, grades 4 and 5 accounted for 31.1%, which was lower than the percentage of grades 4 + 5 at the eye corners (about 60%). A relatively even distribution of forehead wrinkle scores was observed. Table VII shows the degree of consistency between the five-grade scores for the forehead and those for the eye corners determined by each examiner and the consensus score. The kappa values in the evaluation at the eye corner ranged from 0.400 to 0.657 (mean 0.499), indicating fair-to-substantial consistency, and that at the forehead ranged from 0.322 to 0.441 (mean 0.363), indicating fair-to- moderate consistency. The kappa value was slightly lower at the forehead than at the eye corners. Table II Consensus Values of Evaluation by Five General Examiners of Severity of Wrinkles at the Outer Eye Corners of 164 Subjects Wrinkle grade n (%) 1 34 (20.7) 2 11 (6.7) 3 19 (11.6) 4 66 (40.2) 5 34 (20.7) Total 164