Does the eye testing order for visual fields matter?

Abstract

Purpose : Measurement variability in visual field (VF) examinations is likely related to fatigue. Therefore, long-term measurement variability could be worse for the eye examined second compared to the eye examined first; we test this hypothesis using longitudinal VF data collected from clinics.

Methods : Series of 6 VF exams for 7,760 patients were extracted from a database compiled from 4 glaucoma clinics in England. The first VF in each series was discarded to minimize any perimetric learning effect. Individuals were only included (n= 6,894; 89%) if their eyes were always tested in the same order. Of eyes that were always tested in the same order, 99.95% of the time the right eye was tested first. Rates of mean deviation (MD) loss were calculated per eye (dB/y) using least squares regression and the residuals were calculated (predicted – observed) as a surrogate of long-term variability. Absolute errors were then compared between 1^st and 2^nd eye tested. Measurement variability is closely linked to severity of VF loss. Therefore, to correct for any between eye differences we only considered patients with similar VF loss in each eye at baseline (MD difference between eyes < 3 dB; n=4,528; 58%).

Results : Mean (standard deviation) absolute error for all included eyes was 0.74 (0.01) dB. Absolute errors for eyes tested second were greater than those for eyes tested first, by an average of 0.04 dB (P < 0.001, 95% CI: 0.03 - 0.05 dB). (See Fig 1A.) This effect only varied moderately with age (Fig 1B).

Conclusions : During follow-up the VF of the eye examined second will typically accumulate more measurement error than the one tested first; this might be due to fatigue from perimetric testing. The size of the effect is small given overall long-term variability in perimetric testing in clinics. Still, consideration should be given to eye order testing in situations where measurement precision is important, as for example in clinical trials or when clinical management of a patient indicates one eye should be monitored more ‘closely’ than a fellow eye.

This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.

 

Figure 1: A: Distribution (as density curve) of absolute errors including the median, 5^th and 95^th percentiles for the first (red) and second (blue) eye tested (left). B: Running median and 95^th percentile of absolute error with age for the first (red) and second (blue) eye tested (right) – the gap between the lines remains relatively constant against age.