For the purpose of statistical analyses, we excluded participants with a history of corneal disease, ocular trauma, and/or ocular surgery in one or both eyes, as well as participants who had been treated for a prior diagnosis of glaucoma in one or both eyes. These participants were excluded, because either we could not estimate IOP accurately in their eyes (eyes with corneal stromal scars, eyes on treatment for glaucoma) or CCT was likely to be modified by the underlying disease state (e.g., eyes with evidence of Fuchs’ dystrophy, corneal edema, or stromal scarring). For each participant included in the study cohort, we characterized each eye as normal, exhibiting ocular hypertension, or first diagnosed as having glaucoma during the clinic examination (previously untreated glaucoma). Ocular hypertension was defined as IOP of 21 mm Hg or higher, with no evidence of glaucomatous optic nerve damage or glaucomatous visual field loss. Glaucoma was defined by the presence of evidence of characteristic or compatible glaucomatous optic nerve damage and two reliable congruent visual field test results characteristic or compatible with glaucomatous abnormality (after other possible causes for the visual field findings were excluded). Characteristic and/or compatible visual field defects included: nasal steps, paracentral defects, arcuate defects, central islands, temporal islands, and altitudinal loss. Characteristic or compatible glaucomatous optic nerve damage included asymmetric cupping with an interocular difference of 0.3 or more; horizontal or vertical cup-to-disc ratio of 1.0; cup/disc of 0.8 or more and limited loss of neural rim tissue to disc margin, including notching of neural rim tissue, diffuse thinning of neural rim, disc or peripapillary nerve fiber layer hemorrhage, and thinning or defect in the nerve fiber layer in the arcuate areas. IOP was not considered in the diagnosis of glaucoma. Normal eyes had no clinical evidence of glaucoma or ocular hypertension detected in the ophthalmic examination.
We selected one eye of each participant for the statistical analysis, using the following algorithm: If both eyes were normal, then one eye was chosen at random. If only one eye had ocular hypertension or glaucoma, then that eye was chosen for statistical analysis. If both eyes had ocular hypertension or glaucoma was diagnosed in both eyes, then one eye was chosen at random.
Statistical analyses were conducted on data from all selected eyes, using this selection algorithm, and was stratified by the subgroups of normal, ocular hypertensive, and glaucomatous eyes. Normal ranges for CCT were calculated in all eyes and in each subgroup of eyes. In addition, age and gender-specific normal ranges of CCT were calculated in all eyes and in the subgroup of normal eyes. A 95% normal range (defined as the mean ± 1.96 SD) summarizes the range of CCTs in 95% of the eyes in the subgroup of interest (e.g., normal eyes, ocular hypertensive eyes, eyes in participants ≥70 years of age).
The relationship of CCT was contrasted across subgroups using analysis of variance with pair-wise comparisons by the Tukey honest significant different (HSD) procedure. To determine the correlation of CCT with IOP, linear regression and correlational procedures (with and without adjusting for age and gender as covariates) were used. The dependent variable was CCT, and the independent variable was IOP. In addition, the relationship between CCT and IOP was modeled with a multivariate adaptive regression spline (MARS) procedure. All statistical testing was conducted at the 0.05 significance level and was performed on computer (SAS, Cary, NC; MARS, Salford Systems, San Diego, CA).