Abstract
Purpose:
To investigate whether the 24-hour IOP-related pattern contains discriminative information specific to glaucoma.
Methods:
A database containing 24-hour IOP-related patterns (n=306) from 11 clinical studies was analyzed to discriminate between patients with primary open angle glaucoma (POAG; n=219 eyes) and healthy subjects (n=86, not matched for age or gender). The curves were recorded with a contact lens sensor (CLS) that measures spontaneous dimensional changes of the eye at the corneoscleral junction (SENSIMED Triggerfish®, Sensimed, Switzerland). The IOP-related patterns were smoothed using locally weighted polynomial regression (LOESS), with smoothing parameter selected automatically by criterion (Hurvich et al. 1998). For each curve, 13 IOP-related pattern parameters were computed, including variability, number of peaks, wake-to-sleep and sleep-to-wake slopes. The difference between glaucomatous and healthy subjects was assessed by logistic regression on each parameter separately, controlling for age and gender, as well as by a multiple logistic regression model that included several of the parameters and age. Statistics for each parameter included: sensitivity and specificity (selected to maximize the Youden Index) and area under the receiver operating characteristic curve (AUC of ROC).
Results:
For patients with POAG and healthy subjects age was 61.4±13.9 and 43.6±17.7 years (p<0.001), and the percentage of females was 48.3% and 46.6%, respectively. The significant age difference between groups was controlled for the logistic regression, which was significant overall (p<0.001). Of the 13 parameters derived from smoothed IOP-related patterns, 9 were statistically significantly different between the two groups. When adjusting for age and gender, the same 9 out of 13 CLS parameters differed between healthy and glaucomatous eyes. The multiple regression model yielded an AUC of ROC curve 0.84 (p<0.001; 95% CI 0.79-0.89). Using the score generated by the model, sensitivity and specificity were 0.74 and 0.78, respectively.
Conclusions:
Parameters derived from continuous 24-hour IOP-related patterns recorded by a CLS may distinguish between glaucoma and healthy subject, independent of absolute tonometry values. These data and its clinical implications warrant further investigation in clinical studies.
Keywords: 464 clinical (human) or epidemiologic studies: risk factor assessment •
568 intraocular pressure