How Many Visual Fields Are Required to Precisely Predict Future Test Results in Glaucoma Patients When Using...

Area of Science:

• Ophthalmology
• Medical Statistics

Background:

• Accurate prediction of future visual field (VF) sensitivity is crucial for managing glaucoma progression.
• Various regression models exist for predicting VF outcomes, but their efficiency with limited data is debated.

Purpose of the Study:

• To determine the minimum number of VF tests required for precise future VF prediction.
• To compare the predictive accuracy of different regression models, including ordinary least squares linear regression (OLSLR), quadratic, exponential, logistic, and M-estimator robust regression.

Main Methods:

• Analysis of 15 Humphrey Field Analyzer (HFA) 24-2 SITA standard VFs from 247 eyes of 155 open-angle glaucoma patients.
• Prediction of future point-wise (PW) VF results and mean VF sensitivities using varying numbers of VFs within each regression model.

Main Results:

• For PW-OLSLR, 10 VFs were needed to reach the 95% confidence interval of minimum absolute prediction error (2.4 ± 0.9 dB).
• M-estimator robust regression demonstrated significantly smaller absolute prediction error than PW-OLSLR (P < 0.01).
• Exponential and quadratic models showed larger errors; logistic regression errors were not significantly different from PW-OLSLR.

Conclusions:

• Around 10 VF tests are sufficient for accurate prediction of both PW and mean VF sensitivity.
• The M-estimator robust regression model significantly minimizes prediction error for PW VF sensitivity compared to OLSLR.