Can Swedish Interactive Thresholding Algorithm Fast Perimetry Be Used as an Alternative to Goldmann Perimetry in Neuro-ophthalmic Practice? FREE

FORMAL VISUAL field testing is commonly ordered in patients with neuro-ophthalmic disorders such as optic neuropathies or intracranial lesions involving the visual pathways. However, performing visual field tests may be a challenge in patients confined to a wheelchair, those who are unable to communicate, those with cognitive disorders, or patients with severely decreased vision. Manual Goldmann kinetic perimetry (GVF) is classically considered to be the standard perimetry technique for patients with neurological disorders. The technique is easy for the patient, fixation is continuously monitored by a technician, the test can be shortened depending on the patient's alertness, and the technician is able to stimulate a sleepy or poorly cooperative patient as needed. However, not all centers have trained personnel capable of performing reliable GVF testing.^{1 - 2}

During the past 2 decades, many different automated perimetry programs have been developed with the goal of providing a standardized, accurate assessment of the visual field, with less technical expertise required of the operator.^{1 ,3} Most automated perimetry is superior to GVF in terms of sensitivity and quantification of the visual field defects.^{1 ,3 - 7} Automated perimetry has become the standard for visual field testing in glaucoma patients, and it is now widely available.^{1 ,3} However, few studies have evaluated automated perimetry in patients with neuro-ophthalmic diseases, and fewer still have compared GVF with automated perimetry in this group of patients.^{1 ,4 - 11} Most recent studies used the full-threshold Humphrey field analyzer, which is now the most common automated perimeter in the United States, and documented its usefulness in the quantification of neuro-ophthalmic visual field defects.^{9 ,11} They also confirmed, however, that most automated perimetric testing (especially that which uses the full-threshold Humphrey field analyzer) requires a higher level of understanding and greater concentration by the patient, often limiting its use in neurologically impaired patients.^{1 ,5 ,8 - 9 ,11}

A new family of automated perimetry, the Swedish Interactive Threshold Algorithms (SITA), was recently developed to shorten perimetric test time without reducing data quality as compared with the full-threshold Humphrey strategy.¹² The SITA concept allows flexibility of many test parameters. It updates threshold values and adjusts time between stimulus presentation based on the patient's response. The SITA technique has been shown to produce the same high-quality test results as the full-threshold Humphrey strategy in glaucoma patients, but with considerably fewer stimulus exposures.^{12 - 15} Furthermore, the SITA standard strategy has a considerably shorter test time(50%) compared with the full-threshold Humphrey test.¹² SITA Fast is another threshold strategy that is even more rapid than the SITA standard strategy.¹⁶ It is based on the same algorithms, and it has good reproducibility, even though it may be slightly less sensitive than SITA standard in glaucoma patients.^{16 - 20} SITA Fast's shorter test time and flexibility of test parameters would be expected to reduce visual fatigue, thereby improving cooperation in patients with neurogenic visual field defects. Indeed, based on the patient's response, the SITA strategy continuously updates threshold values during the test, and it automatically adjusts the time between stimulus presentations in response to the reaction time of the patient. In this study, we compared SITA Fast computerized static perimetry with GVF in the detection and characterization of visual field defects in neuro-ophthalmic practice.

We prospectively evaluated consecutive patients seen with either severe neurological impairment or severe vision loss in the Neuro-Ophthalmology Unit at Emory University (Atlanta, Ga) between September 2000 and April 2001 (Table 1 and Table 2). Severe neurological impairment was defined by a score of 3 or 4 on the Modified Rankin Scale (MRS) (MRS 3 = moderate disability: requires some help, but able to walk without assistance; MRS 4 = patient unable to walk: requires permanent help).²¹ Severe vision loss was defined by a visual acuity of 20/200 or worse in at least one eye. Patient inclusion criteria consisted of age 18 years or older, the ability to understand instructions, and the motor ability to carry out a visual field examination (patient able to sit upright for at least half an hour and to press a button in response to visual stimulation). Patients not willing to have both GVF and SITA Fast perimetry on the same day were excluded.

Visual field examinations were performed using the GVF and the Humphrey automated static perimeter with the SITA Fast algorithm. Both tests on both eyes were always performed on the same day, with the GVF examination performed first. The GVF was performed by the same skilled technician. Patients were seated before the Goldmann perimeter with the left eye occluded first. Each patient's near refraction, with additional diopters adjusted for age, was provided. The machine was calibrated according to the manufacturer's instructions, and the background-target luminosity ratio was set at 1:33. The blind spot was mapped using the I2e or I4e test object (depending on the patient's visual acuity) at a distance of 300 mm to ensure patient reliability. Relative defects in the visual field were detected by using standard test objects such as V4e, I4e, I2e, I1e, with additional isopters plotted as indicated. To mark the peripheral edge of an isopter, the test object was moved at a rate of 2° to 3° per second from the far periphery toward fixation until it was seen. For scotoma testing, the test object was presented inside the region of field loss and moved radially in a straight line until it was seen. The left eye was then tested in the same fashion.

SITA Fast perimetry was obtained for all patients after at least 1 hour of rest. We used a Humphrey 740 perimeter with the standard settings of a size III (4 mm²) test object at a distance of 333 mm, with a 200-millisecond stimulus duration, and a bowl illumination of 31.5 apostilb (asb) as previously described.^{12 ,16} To perform the fastest test, we chose the 24-2 strategy (exploring the central 24°) rather than the 30-2 strategy. Each patient's fixation and position were checked every 1 to 2 minutes on the video eye monitor, with adjustments made as necessary. The right eye was tested prior to the left eye.

The GVF was considered unreliable if the technician performing the test assessed the patient's cooperation and fixation to be too poor to plot an adequate field, or if the blind spot could not be plotted. A SITA Fast visual field was considered unreliable if fixation losses were 50% or more. We did not use false-positive and false-negative catch trials.

The 3 investigators made an independent subjective assessment of the pattern configuration, extent and depth of the visual field defects on the hand-drawn Goldmann chart, and on the pattern SD and the graytone printout from the SITA Fast perimeter. Direct comparison was made between the central 24° of the GVF as assessed by putting a template over that area, and with the pattern SD and the graytone printout from SITA Fast (Figure 1).

The results of the visual field comparison were classified into 1 of 9 groups, as previously suggested by others (Table 3).^{5 ,9}

The testing time required for each visual field strategy in each eye was compared using the χ² test. The patient's functional status was assessed with the MRS and the Barthel Index,²¹ on the day of the visual field tests. Patient preference was evaluated by asking the patient which visual field test they would rather have on their follow-up examination.

A total of 64 patients was included in the study. There were 36 men and 28 women with a mean age of 53 years (range, 18-92 years). Patients were divided into 2 groups, depending on their neurological status and visual acuity. Twenty-five patients (17 men, 8 women; mean age, 51 years [range, 18-86 years]) with severe neurological impairment (MRS 3 or 4) were included in the first group (Table 1). All 25 patients were able to perform both visual field tests with both eyes, and all 50 eyes were included in the analysis. The mean MRS was 3.4 (range, 3-4), and the mean Barthel index was 52.4 (range, 25-85). Five patients with neurological deficits also had 8 eyes with poor visual acuity, ranging between 20/200 and hand motions. The other 42 eyes had a mean visual acuity of 20/30 (range, 20/20-20/100). Thirty-nine patients with severe vision loss (19 men, 20 women; mean age, 54 years [range, 18-92 years]) were included in the second group(Table 2). Among these 39 patients(representing 78 eyes), 3 patients had 1 eye with a visual acuity of no light perception, and 25 patients had 1 eye with a visual acuity better than 20/200. These 28 eyes were excluded from the study, and the analysis was performed on the remaining 50 eyes. Visual acuity was extremely poor (20/400 or worse) in 34 of 50 eyes.

Clinical characteristics of the patients and visual field description, reliability, test time, and categorization of the visual field comparison for each of the 100 eyes included in the study, are detailed in Table 1 and Table 2.

Visual field examinations with GVF were reliable in 77% of all eyes. Visual fields obtained with the SITA Fast strategy were also estimated to be reliable in 77% of all eyes. Among the 50 eyes of patients with severe neurological deficits, 32 (64%) had a reliable GVF, and 36 (72%) had a reliable SITA Fast visual field. Among the 50 eyes with severe vision loss, 45 (90%) had a reliable GVF, and 41 eyes (82%) had a reliable SITA Fast visual field. In 16% of all eyes (7 [14%] of 50 eyes in patients with severe neurological deficits, and 9 [18%] of 50 eyes with severe vision loss), the GVF was reliable, but the SITA Fast was not. In 14% of all eyes, (11 [22%] of 50 eyes of patients with severe neurological deficits, and 3 [6%] of 50 eyes with severe vision loss), the SITA Fast was reliable, but the GVF was not. In 7 (14%) of 50 eyes of patients with neurological deficits, neither of the visual field tests were reliable.

The distribution of the results of the visual field comparisons is demonstrated in Figure 2. Overall, the 2 fields were similar (groups 1, 2, 3, and 4) in 75% of all eyes. Among the eyes of patients with neurological deficits, 35 (70%) of 50 eyes had similar visual field tests on both strategies (Figure 3).However, 11 (22%) of 50 eyes of patients with neurological deficits had normal visual fields (group 4) (Figure 2). Excluding these 11 healthy eyes from the analysis, 24 (61.5%) of 39 eyes of patients with neurological deficits had similar visual field defects with both tests. Among the eyes with vision loss, 40 (80%) of 50 had similar visual field defects on both fields (Figure 4).A few eyes were classified in groups 5 and 6, in which both visual field tests were reliable, with one of the tests being abnormal and the other one being healthy. In 8% of all eyes (6 of 43 eyes of patients with neurological deficits, and 2 of 50 eyes with vision loss), GVF failed to show a defect demonstrated by SITA Fast (group 7) (Figure 5). In 9% of all eyes (3 of 43 eyes of patients with neurological deficits, and 6 of 50 eyes with vision loss), SITA Fast failed to show the visual field changes demonstrated by GVF (group 8). In 4 of these patients categorized as group 8, the visual field defect or the residual island of vision was located at the edge or outside of the central 24° explored by automated perimetry(Figure 6).

The mean ± SD test time on the GVF perimeter was 7.97 ± 3.2 minutes per eye (range, 3-22 minutes). It was 8.04 ± 3.57 minutes(range, 4-22 minutes) in the group of patients with neurological deficits, and 7.9 ± 2.7 minutes (range, 3-15 minutes) in the group of patients with poor vision. The mean ± SD test time on the SITA Fast perimeter was 5.43 ± 1.41 minutes per eye (range, 3.03-11.4 minutes). It was 5.44 ± 1.54 minutes (range, 3.05-10.52 minutes) in the group of patients with neurological deficits, and 5.79 ± 1.30 minutes (range, 3.40-11.73 minutes) in the group of patients with poor vision. The SITA Fast perimeter reduced the test time by 2.54 minutes (47%) compared with the GVF perimeter(P<.001). The amount of reduction of test time with the SITA Fast perimeter was similar in the group of patients with neurological deficits and in the group of patients with severe vision loss.

When asked which visual field test they would rather have on their follow-up examination, 58 (91%) of our 64 patients preferred GVF. The main reason given by the patients who preferred GVF was the difficulty of maintaining concentration during testing with the SITA Fast. Among the 6 patients who preferred SITA Fast, one had severe neurologic deficits, and 5 had severe vision loss. They all enjoyed the computerized aspect of the test.

Goldmann perimetry is the traditional method for evaluating visual field defects in patients with severe neurological handicaps or severe vision loss. With the development of more sophisticated, reliable, sensitive, affordable, and easily performed automated perimetry programs, GVF performed by a skilled technician has become less available. However, although the full-threshold Humphrey analyzer has proven to be one of the most sensitive and reliable automated perimetry strategies for more than 20 years, it is still only rarely used by neurologists. Indeed, it does have drawbacks compared with GVF, such as prolonged test time, the rapid appearance and disappearance of the light stimulus, lack of human contact and reassurance, and continued testing despite detection of poor fixation.¹ Previous studies^{8 - 9} have shown that the full-threshold Humphrey analyzer cannot overcome many of the major obstacles to accurate visual field assessment, such as fixation losses, poor concentration, and patient fatigue, which are all common findings in neurologically disabled patients. The SITA family of automated perimetry uses the Humphrey perimeter with different algorithms, making the visual field testing process much shorter and easier for the patient.^{1 ,12 ,16} These automated perimetry programs have replaced the full-threshold Humphrey analyzer in most glaucoma centers and will soon be readily available and accessible in most ophthalmic and neuro-ophthalmic practices.

Our study shows that SITA Fast computerized static perimetry, a new rapid perimetric threshold test, can be used to identify and localize visual field defects in most patients with neuro-ophthalmic diseases. Previous studies^{16 - 17} have shown that SITA Fast is reliable in healthy subjects and in glaucoma patients, in whom visual acuity is usually relatively well preserved. We evaluated only patients with either severe vision loss who may not be able to see the standard target used on the automated perimeter, or those with a neurologic deficit that may compromise their ability to perform a computer-driven test. We assumed from previous studies^{5 ,8 - 9 ,11 ,17} that patients with good visual acuity or mild neurological deficits would not have trouble performing SITA Fast perimetry, and therefore, we excluded such patients from our study.

The overall reliability of visual field testing in our study seemed to be very good (77%) even in our disabled patient population. However, without an accepted established definition of a "reliable visual field test" for both GVF and SITA Fast perimetry, these results should be interpreted with caution. Our estimation of a "reliable" GVF was based on the technician's subjective assessment of the patient's cooperation. For SITA Fast perimetry, we used a very high rate of fixation loss (>50% rather than >20%) to establish that a visual field was not reliable. Sanabria et al²² showed that fixation losses result not only from subjects looking around, but also from faulty initial localization of the blind spot and, therefore, that these losses may be the result of technical artifacts. Using our criteria, we observed similar reliabilities with both GVF and SITA Fast perimetry in the group of patients with poor visual acuity. For the SITA Fast strategy, we used the standard size III target provided by the Humphrey analyzer, which corresponds to a 4-mm² stimulus size (equivalent to a size III target on GVF). This target allowed reliable evaluation of the visual field of patients with visual acuities as poor as hand motions. Nevertheless, in 9 eyes with vision loss, and in 7 eyes of patients with neurological deficits, GVF was reliable, but SITA Fast was not, according to the high percentage of fixation losses. In this group of patients who had an unreliable SITA visual field, most patients with vision loss had visual acuities worse than 20/400 or were older than 72 years. It is likely that a larger stimulus (such as 64 mm², equivalent to the size V target on GVF) would have provided more useful information in these patients.²³ However, we selected the "standard" stimulus size provided by the standard Humphrey package, as used by most community-based ophthalmologists, to simulate common referral conditions. We felt it would be too taxing on the patients who failed to perform a reliable SITA test to repeat perimetry with a larger stimulus on the same day. Most patients with neurological deficits who were not able to perform a reliable SITA Fast had either a cerebellar syndrome compromising their coordination and their fixation, or frontal or occipital lesions associated with spatial and cognitive disorders (Figure 7). Without the experience of the highly skilled technician who performed all GVFs, it is likely that most of these patients would not have been able to perform a reliable GVF. Nevertheless, in 22% of our patients with neurological deficits, SITA was more reliable and provided better visual field information than GVF (Figure 5). This may be explained by the short examination time of the 24-2 SITA Fast perimetry, as well as the flexibility of the SITA Fast parameters.¹⁶

The results obtained with SITA Fast perimetry indicate a relatively good correlation with Goldmann perimetry in the detection, characterization, and quantification of visual field defects in this particular population. We found that 75% of all eyes with abnormal visual fields had similar visual field results with GVF and SITA Fast (70% of eyes in patients with neurological deficits, and 80% of eyes with severe vision loss). Only a few previous studies have compared GVF with automated perimetry in patients with neuro-ophthalmic disorders. Various automated perimetric strategies were used, including the Fieldmaster,^{7 - 8} the Octopus,⁵ and the Humphrey full-threshold analyzer.^{9 - 11} These studies showed that automated perimetry was comparable to GVF in detecting visual field abnormalities in neurologic diseases. For example, visual field defects were almost identical with automated perimetry and GVF in 84% of the 25 patients studied by McCrary and Feigon,⁵ and in 87% of the 69 eyes studied by Beck et al.⁹ The purpose of these studies was to evaluate the reliability of automated perimetry in identifying and quantifying visual field defects from neuro-ophthalmic diseases such as nonglaucomatous optic neuropathies and lesions involving the retrochiasmal visual pathways. None of these studies specifically addressed the issue of severely decreased vision in some patients with optic neuropathies, nor did they correlate their results to the degree of a neurological handicap. Most of these studies^{5 ,7 - 9} used older-generation perimeters that are much slower and generate more visual fatigue than the SITA Fast algorithm used in our study.

A quarter of all eyes, representing 25 (28%) of 89 eyes with abnormal visual fields, were categorized into group 2, in which SITA Fast showed a slightly larger visual field defect than GVF (Figure 4). It has been shown that automated perimetry is more sensitive than GVF in the detection of visual field defects in patients with glaucoma.^{4 ,6 ,8 - 9} Indeed, even in the hands of a skilled operator, GVF often underestimates the severity of the visual field defects, especially when the defects are located in the central part of the visual field.^{4 ,6 ,8 - 9} Another possible explanation is statokinetic dissociation, which has been reported in various pathological cases involving the optic nerves as well as the occipital lobes.^{24 - 25} Statokinetic dissociation is a physiologic phenomenon related to the easier perception of moving objects (as in Goldmann kinetic perimetry) than stable objects (as in static automated perimetry), giving rise to a greater visual field defect on automated perimetry compared with GVF.^{24 - 25} However, in 5 eyes with severe vision loss, GVF showed a slightly larger defect than SITA Fast (group 3).

In 9 (10%) of 89 eyes with abnormal visual fields, SITA Fast failed to show a visual field defect that was demonstrated by GVF (group 8). In 4 of these eyes, the visual field defects (or the residual island of vision) were localized at the border or outside of the central 24° of visual field evaluated by SITA Fast (Figure 6). Goldmann perimetry tests the entire visual field (180°) and is hence the obvious technique of choice in eyes with either residual eccentric islands of vision or with visual field defects not involving the central 24° of vision. The SITA software allows for the evaluation of the central 10°, 24°, or 30° of vision. We used the 24-2 instead of the 30-2 strategy to reduce the duration of the test, thereby limiting visual fatigue. However, it is unlikely that evaluation of the central 30° would have changed our results.²⁶ The full-threshold Humphrey analyzer is able to test the central 60° of vision, but the length of the test(as long as 30 minutes per eye) precludes easy usage. Development of a SITA strategy testing the central 60° of visual field could potentially solve this problem, with an acceptable minimal increase in test duration within the SITA Fast algorithm. However, it is also possible that certain lesions involving the retrochiasmal visual pathways may not be detected as well with automated perimetry as with GVF, even if the defects fall within the central visual field. For example, one of our patients with bilateral occipital infarctions had a GVF perimetry that exquisitely delineated bilateral homonymous defects—findings completely missed by SITA Fast perimetry, which was unreliable (Figure 7). Wong and Sharpe¹¹ evaluated 12 patients with occipital lobe infarctions using tangent screen, GVF, and the full-threshold Humphrey visual field, and correlated the findings with magnetic resonance imaging of the causative lesions. They observed that even though Humphrey automated perimetry was able to detect the visual field defects, it incorrectly localized the defects to the proximal portion of the retrochiasmal pathway in 2 patients, failed to detect sparing of the occipital pole in 4 patients, and overestimated the lesion size in 1 patient.¹¹

The duration of visual field testing was significantly shorter for SITA Fast than for GVF. Considering that SITA adjusts the time between stimuli based on the patient's answers, and that our experienced Goldmann perimetrist is faster than most, this difference is impressive. It helps explain why even patients with cognitive disorders and poor concentration were able to reliably perform a SITA Fast visual field test. Although the GVF was consistently performed first in all patients, the GVF and SITA Fast techniques are extremely different; it is therefore unlikely that a learning effect can explain the discrepancy between the 2 visual fields observed in some patients, or the shorter SITA Fast test duration.

Similar to previous studies,^{5 - 6} we found that nearly all our patients preferred GVF to SITA Fast perimetry. Our patients noted that it was difficult to maintain concentration without some communication with the examiner, and that the standard size III object was hard to see on the SITA Fast. The 6 patients who preferred SITA Fast were, in general, younger patients who seemed to enjoy the computerized method.

Our results suggest that SITA Fast perimetry could be ordered instead of GVF in most patients with optic neuropathies or lesions involving the intracranial visual pathways. Additionally, these findings may be applicable to younger children, in whom the realization of a reliable visual field is often challenging.¹ However, GVF may still be the test of choice in patients with occipital lesions, in those with peripheral visual field defects, and in those with large central defects of more than 30°. Furthermore, it is likely that GVF performed by a skilled operator is preferable to SITA Fast in patients with suspected nonorganic vision loss.

In conclusion, we believe that the SITA Fast strategy of automated perimetry may be useful in the evaluation of central visual field defects associated with neuro-ophthalmic disorders. The development of additional SITA software that could test out to 60° might allow even better use of SITA strategies in neuro-ophthalmic practice. Our results suggest that for the general ophthalmologist or neurologist, visual field testing with SITA Fast perimetry might even be preferable to GVF (especially if the GVF is performed by a marginally trained technician) even in patients with severely decreased vision or who are neurologically disabled.

Submitted for publication October 11, 2001; final revision received February 19, 2002; accepted April 30, 2002.

This study was supported in part by a departmental grant from Research to Prevent Blindness Inc, New York, NY (Department of Ophthalmology). Dr Newman is a recipient of a Research to Prevent Blindness Lew R. Wasserman Merit Award.

Presented at the 53rd Annual Meeting of the American Academy of Neurology, Philadelphia, Pa, May 7, 2001.

We would like to thank Kathy B. Moore, OA, who performed Goldmann perimetry on all the patients.

Corresponding author: Valérie Biousse, MD, Neuro-Ophthalmology Unit, Emory Eye Center, 1365-B Clifton Rd NE, Atlanta, GA 30322 (e-mail: vbiouss@emory.edu).