A comparison of visual field testing with a new automated perimeter, the Compass visual field analyser, and the Humphrey visual field analyser

Abstract

PurposeTo compare a new visual field analyser, Compass, that included an eye tracking and scanning ophthalmoloscopy to Humphrey visual field analyser (HFA).MethodsProspective cross study design.Patients were included after a complete examination: all were indemn of ocular disease except glaucoma. Visual acuity was 20/20 for each eye and spherical equivalent ranged from +3 to −3 D.Patients were randomly assigned to one instrument. HFA was performed with a 24‐2 SITA standard strategy comparable to the Compass 24‐2 ZEST strategy. Both eyes were tested and 30 min after where screened with the other instrument.MD, PSD and exam duration were compared for both devices using a Wilcoxon signed‐rank test. Agreement was evaluated with a Bland‐Altman graph for each parameter.ResultsThe study included 67 eyes of 30 patients. This population was decomposed as: 13 normal eyes, 28 OHT or glaucoma suspect and 26 glaucoma. Patients’ demographics were (mean ± SD): age 66.3 ± 13.0 years, pachymetry 527.6 ± 28.95 μm, axial length 23.73 ± 0.99 mm, spherical equivalent −0.004 ± 1.3 D. 65 HFA vs. 62 compass visual fields were reliable (ns). Mean Deviation was equivalent for HFA and Compass instruments: −1.6 ± 2.6 vs. −1.4 ± 2.8 dB (p = 0.28). Pattern Standard Deviation was significantly higher for the Compass 3.9 ± 2.4 vs. 2.4 ± 1.9 dB for the HFA (p < 0.0001). Examen duration was also longer for the Compass 351 ± 83 s vs. 318 ± 48 s for the HFA (p = 0.0164). Bland Altman plots showed a good agreement between HFA and Compass.ConclusionsThis study shows that MD and failure rate were comparable between both instruments, PSD and examination duration were slightly higher for the Compass. Agreement was good between both instruments.As most of the patients included had already performed a HFA before, this may explain the duration difference between and should be consider in further explorations.

Compass fundus automated perimetry

“…After its clinical introduction in 2014, CMP had received validation against the clinical standard for perimetry, that is, HFA. Despite a low number of publications, analyses on both normal and glaucoma subjects agree on the fact that the two instruments are interchangeable, 11,19,22,23,29 as the difference of their global indices (MD, GSS2, VFI) is largely inferior than test-retest variability for HFA. 9,18 Although one paper claim that the results of the two instruments are hardly interchangeable due to large differences in peripheral sensitivities, such study seemed not to consider adequately the physiological variability of perimetry 9,18 and it was also flawed by several limitations of methods.…”

“…A 24-2 grid was evaluated with ZEST in three studies, [22][23][24] All of them confirmed the improvement in test duration. One study compared ZEST versus FT on 33 subjects (18 normal and 15 glaucoma subjects) demonstrating that mean test duration nearly halved compared to FT: 353 ± 31 s for normal subjects and 343 ± 67 s for glaucoma patients, respectively, corresponding to a mean time of 6.4 and 6.2 s/location.…”

“…As for FT, PSD was significantly higher for CMP 3.9 ± 2.4 dB vs 2.4 ± 1.9 dB for HFA (p < 0.0001). 23 These findings were partially challenged by Rao et al 24 In their paper, they evaluated 97 eyes of 58 subjects and excluded 46 eyes of 22 patients due to unreliability; data of 51 eyes of 36 subjects (33 eyes with glaucoma and 18 with glaucoma suspect) were finally available for comparison (see section 'CMP reliability' for reliability). In this group, MS was significantly lower with CMP: median was 27.2 for CMP (interquartile range: 25.4; 28.5 dB) and 28.1 for HFA (25.6; 30.0 dB, p = 0.04).…”

“…[26][27][28] The reliability of results between CMP and HFA was compared in three studies. [22][23][24] In the first study, both inexperienced and experienced patients received the two tests in randomized sequence. For inexperienced subjects, a demo was performed just with the first perimeter to be tested (a specific training for CMP was not applied in all cases).…”

“…Despite statistical equivalence, the authors pointed out how perimetric learning may affect test reliability and hence recommended to rule out CMP-specific learning effect. 23 In the last study, the authors used a definition of reliability which is not standard (fixation loss <20%, false positive <20%, false negative <20%), and they found out that unreliable tests were 38% with CMP and 18% with HFA. The number of unreliable tests due to false negative >20% was higher with CMP (n = 27) than with HFA (n = 3, p = 0.005).…”