Accelerating visual field loss with age: A statistical approach using a large-scale real-world dataset

Abstract

Purpose

To examine whether visual field (VF) loss progresses linearly or accelerates with aging.

Design

Retrospective database analysis study.

Methods

A total of 27,556 VF tests represented 1721 eyes of 973 Japanese subjects who recorded central 30–2 VF at least 10 occasions for each eye. A mixed-effects model with a two-level random-effects structure was employed, where Patient ID served as the first level and left/right eye (LR) nested within Patient ID served as the second level. Using this framework, three models were developed to assess the effects of aging on mean deviation (MD): (1) a linear model that included only an age term, (2) a quadratic model that extended the linear model by adding an age-squared term to capture potential non-linear effects of aging, and (3) an interaction model that incorporated both an age term and a term for the time from baseline in years (t) to evaluate the interaction between age and elapsed time. These models were used to estimate MD slope (dB/year) and acceleration (dB/year²).

Results

In the linear model, the mean deviation (MD) declined at an average rate of approximately −0.17 dB/year (AIC=122,418, BIC=122,459). The quadratic model indicated that VF performance peaked around age 37 and then declined at an accelerating pace (AIC=121,680, BIC=121,729). Among the tested approaches, the interaction model, which incorporated both age and time from baseline, yielded the lowest AIC (119,498) and BIC (119,555), indicating the best fit. These findings collectively suggest that VF decline in glaucoma is not uniform but intensifies with advancing age.

Conclusions

VF loss does not progress at a constant rate but accelerates with aging. Considering a future acceleration of VF loss, in addition to a current speed, is important for lifelong maintenance of visual function in glaucoma patients.