Data were evaluated with commercial software (SPSS, ver. 15.0.1; SPSS Inc, Chicago, IL, and MedCalc, ver. 9.2; MedCalc Software, Mariakerke, Belgium). The Shapiro-Wilk test was used to assess the normality of the quantitative variables. Normally distributed variables were summarized by using the mean and SD, and nonnormally distributed variables were summarized by using the median and interquartile ranges (IR; percentiles 25 and 75). The association between Cirrus and Stratus OCT was tested by determining Spearman's correlation coefficient (
r s). The quantitative variables between two independent groups (the sexes and normal and glaucomatous groups) were compared by using the Mann-Whitney U test nonnormally distributed variables (all but age), and two-sample Student's
t-test was used for normally distributed variables (age only). To compare qualitative variables between independent groups, χ
2 tests were used. RNFL sensitivities and specificities were calculated for both OCTs compared with visual field classification. Receiver operating characteristic (ROC) curves were used to determine the discriminatory capabilities between healthy and glaucomatous eyes. The areas under the ROC curves (AUC) were compared by using the Hanley-McNeil method for paired data.
6 A difference between AUCs of 0.1 was considered to be clinically relevant. Using statistical software (Medcalc), a sample size of 57 eyes per group (control/glaucomatous) was estimated to detect a difference in AUC over 0.1 with a power of 95% at a significance level of 5%; for a power of 90%, a sample of 48 was needed. The sensitivities for 85% and 95% fixed specificities were calculated. Diagnostic classification of both OCT (outside normal limits, borderline, or within normal limits) provided by each instrument after comparison with its respective normative database was also evaluated, and likelihood ratios (LRs) were reported. The agreement of the global RNFL values (quantitative data) was evaluated by using the Bland-Altman method.
7 The agreement of the RNFL classification (qualitative data) between both OCTs was assessed using the weighted kappa (κ) coefficient.
8 The strength of the qualitative agreement was categorized according to the method of Landis and Koch
9 : less than 0, no agreement; 0 to 0.20, poor; 0.21 to 0.40, low; 0.41 to 0.60, moderate; 0.61 to 0.80, substantial; and 0.81 to 1.00, almost perfect.