This retrospective study found that the new software for the Humphrey Visual Field Analyzer accurately predicts progression in patients with mild to moderate glaucoma, but it is far more accurate when the predicted visual field index (VFI) is at least 90 percent.
Subjects were 42 patients with at least 11 years’ follow-up who underwent annual automated visual field testing. Patients with mean deviations ≤ −20 dB were excluded, and all unreliable results were omitted (fixation losses ≥ 20%, false-positive ≥ 15%, OR? false-negative ≥ 33%). The new software was used to analyze the data from the first half of the series to generate a prediction that was then compared to the actual outcome as assessed during the last five years.
The authors found that even with the addition of selective laser trabeculoplasty in some subjects, the VFI was accurate in predicting the rate of glaucoma progression. There was a slight, although not statistically significant, difference between the predicted and observed values for the 61 eyes included in the study. Progression was overestimated, on average, by 1.37 percent, with 95 percent of the predicted values found to be between 12 percent lower and 9.3 percent higher than the observed values.
The prediction was more accurate if either the estimated slope of the prediction was greater than –1.5 percent/year, with 95 percent of the predicted values within −7.2 percent and +7.2 percent, or the predicted VFI was ≥ 90 percent, with 95 percent of the predicted values within −4.5 percent and +5.2 percent of the actual observed VFI. This is twice as accurate as previously measured with all of the data sets included. However, if the slope was more negative than −1.5 percent per year or the predicted VFI was < 90 percent, the accuracy declined considerably, with 95 percent of the predicted values falling within −13.8 percent and +20.5 percent of the actual value.
The authors note that there is no perfect method to predict visual field progression accurately in all patients with glaucoma. It is important to remember that many variables confound the results generated by computer software, especially when human behavior is involved. Patients' adherence to medication, clinic attendance, visual field testing performance and patterns of progression cannot be predicted or standardized. Nonetheless, for many patients, it is a helpful tool for monitoring glaucoma progression and is proving to be a powerful visual aid and reference for patients to better understand the states and changes of their visual field over time.