Baseline Prognostic Factors Predict Rapid Progression

Abstract

 

Purpose

 

To investigate baseline prognostic factors which predict rapid progression of the visual field (VF) in primary open angle glaucoma patients.

 

Methods

 

767 eyes of 566 subjects from the Advanced Glaucoma Intervention Study (AGIS) and the University of California at Los Angeles’ (UCLA) Jules Stein Eye Institute with primary open angle glaucoma were included. The VF rate of decay for each VF test location was calculated with point-wise exponential regression (PER) analysis and separated into fast and slow components. Subjects with a fast component decay rate of 36%/year or faster were designated as rapid progressors. See Table 1 for prognostic factors assessed in the multiple regression model. For cross-validation of our logistic model, we developed a sampling group containing 2/3 of the entire study group and identified the rapid progressors to which we fitted the same logistic model. We obtained the coefficient estimates and applied them to the 1/3 test group. We calculated the probability of being a rapid progressor for each eye in the test group and calculated the area under curve (AUC) of the receiver operating characteristic curve (ROC). After repeating this procedure 1,000 times, we calculated the average AUC as cross-validation for the model.

 

Results

 

222 eyes were identified as rapid progressors. The average (±standard deviation) age was 65.93(±9.9) years in the rapid progressors. Subjects with worse baseline MD (P<0.0001; odds ratio[OR], 1.12; 95% confidence interval[CI], 1.09 to 1.16), greater vertical C/D ratio at baseline (P=0.001; OR, 1.23; 95% CI, 1.09 to 1.39), and older age at baseline (P=0.030; OR, 1.23; 95% CI, 1.02 to 1.49) were more likely to have a fast VF component rate of 36%/year or faster than did subjects without these characteristics.

 

Conclusions

 

Worse baseline MD had the highest association for classifying the rapid progressors as defined by the VF fast component worsening by 36%/year. Both increasing baseline vertical C/D ratio and increasing age at baseline were also associated with this classification.

 

 

Table 1. Results of multiple logistic regression using 11 the variables

 

 

Figure 1. The plot was the ROC curve of predicting rapid progressors with a multiple logistic regression model. The sensitivity and specificity were calculated based on the predicted probability of the logistic model. Based on 1,000 AUC values, the predictive value of the logistic model was 0.72.

 

Keywords: 464 clinical (human) or epidemiologic studies: risk factor assessment • 758 visual fields