Point-wise visual field sensitivity analysis can be improved by incorporating an overall visit effect into the data analysis, according to a study by researchers from the Rotterdam Eye Institute.
Rotterdam, Netherlands-Point-wise visual field sensitivity analysis can be improved by incorporating an overall visit effect into the data analysis, according to a study by researchers from the Rotterdam Eye Institute.
Designated as the Global Visit Effects (GVE), this new parameter aims to incorporate both measurable factors-such as season, time of day, and reliability indices-and unknown transient factors, such as fatigue, lack of concentration, or delayed reaction time, said Susan R. Bryan, MSc, researcher and PhD candidate, The Rotterdam Ophthalmic Institute, The Netherlands.
“One of the difficulties in modeling visual field data is the large measurement variability, partially due to the subjective nature of the device,” Bryan said. “Random measurement errors that are present in the point-wise sensitivity estimates are reduced when calculating summary parameters such as the mean deviation. Other errors, however, are spatially correlated and affect the whole visual field.”
The GVE contribution was investigated in a study using data from 50 patients with primary glaucoma who underwent 24-2 full threshold testing.
The analyses included sensitivity estimates in 52 locations in both eyes and used a hierarchical, Bayesian, mixed-effects model with four levels: individual, eye, hemisphere and location.
When comparing the observed data and the posterior predicted values for the models, it was determined that the GVE significantly improved the model fit, and mean absolute error was 1.94 dB with the GVE and 2.14 dB without it, Bryan explained.
Moreover, the magnitude of the GVE effect, 0.82 dB, was much larger than that of any other measurable factor known to affect visual field testing, including season, time of day, percentage of fixation losses, percentage of false positives, and percentage of false negatives, and was 3-fold larger than that of all the measurable factors combined.
“Measurable factors-such as season, time of day and reliability indices-represent one group of such errors and have been evaluated before,” Bryan said. “Although these factors are statistically significant, as our study showed, they are rather small, and as such cannot explain the observed global variation in visual fields.
“Our findings confirm our hypothesis that other transient factors modeled as a global visit effects play a more important role, and showed that by incorporating the GVE in our models, better estimates of the real evolution in visual field data over time may be obtained,” she continued.
In addition, Bryan said including the GVE changes the estimation of the rate of visual field progression, and also affects the projected future visual field sensitivities, and may therefore have an impact on treatment.
For now, clinicians should be aware that even if a new measurement is outside the range of previous measurements, it might just be due to the GVE rather than actual progression, Bryan commented.
“Additional visual fields or other clinical sources of information should be used to confirm the change,” she recommended.