Abstract
Purpose.:
To introduce a novel approach to topographic function assessment in visual impairment that requires neither fixation nor reading.
Methods.:
One hundred thirty-five consecutive low vision patients with varying diagnoses and 30 control subjects of comparable median age participated. Performance was measured in a search task that required finding and identifying visual targets which appeared consecutively on a monitor in 32 locations of the central field of gaze. The task specifically discourages steady fixation and the subjects could make eye movements as needed to locate targets. Target size was always double the size threshold, and no manual action was required. The best attainable reading speed at any size was routinely measured (MNread). Main outcome measure was response latency necessary to solve the task. Data were median latencies and sums of all latencies.
Results.:
Measurements yielded a wide variety of performance levels, with a factor of 14 to 16 between best and worst performers. The highest correlation existed between median response latency in the search task and best attainable reading speed. Only a weak correlation was found between performance and visual acuity. No statistically significant correlations were found with age or diagnosis.
Conclusions.:
The “search-and-identify” paradigm and continuous text reading share an important mechanism that determines performance in both tasks. The authors hypothesize that the factor enabling patients to perform well in both paradigms is oculomotor skill and/or eye movement strategy. Results show that the search test is a useful tool for the easy assessment of impaired vision independent of language, level of literacy, and reading habits.
Visual impairment can be characterized as diminished functional vision.
1 This characterization cannot be achieved by measuring visual functions like visual acuity alone, although it is often used as a substitute estimate of functional vision. Characterizing visual impairment needs to include an assessment of the patient's ability to interact with the visual environment. The latter should include the capability to learn, to remember, and to respond to changing stimulus conditions. The most important response of this kind are eye movements, which are elicited by attracting visual attention to a presumably interesting target on the peripheral retina.
2 –4 It has been reported that patients with low vision can perform feature search.
5 In vision rehabilitation, it has also been shown that search performance can be trained.
6 Being able to make eye movements can decide whether training is effective or not.
7 The process may involve the adoption of new eye movement strategies.
8 –10 This article aims to introduce a novel approach toward the assessment of visual impairment by removing some traditional barriers.
A realistic way to measure visual impairment calls for topographic testing without the constraint of strict fixation. The history of topographic vision assessment has been dominated by the necessity to bring individual responses into spatial register to preserve spatial relations in the visual field. This is done by requiring stable central fixation. This requirement has been an essential issue in perimetry since its development in the mid-19th century (for its history, see
http://webeye.ophth.uiowa.edu/ips/PerimetryHistory). However, steady fixation is hard to accomplish for most patients with low vision, especially those with central scotomas and damaged foveal vision. Consequently, these patients may use a preferred retinal locus (PRL) to accomplish the goal,
11,12 which is known to be less accurate than in subjects with an intact fovea.
13 –15 Vision assessment can be made easier by using the physiological blind spot in conventional perimetry as reference scotoma,
16 or by offering the patient abundant peripheral landmarks to enable them to monitor their own fixational stability.
17 –19 Nonetheless, unstable fixation during topographic vision assessment has remained a problem that can be overcome only by an instrument that does not require calibration (i.e., a scanning laser ophthalmoscope [SLO]),
20 especially if it can compensate for fixational inaccuracies.
21 But what do you do if you have no SLO available?
A second important issue is the fact that the task in perimetry is to simply detect the appearance and disappearance of a target during steady fixation. In real life, our eyes are always moving, and the movement can effectively minimize the impact of field disruptions on performance of visually-guided activities, such as reading. To differentiate perceptual deficits in patients with low vision more finely, it is desirable to use a task that is similar to a real-life viewing situation and that makes the test more sensitive by making the task more demanding. A simple step is to require solving a discrimination task, where a positive identification is necessary.
17,22,23 Increasing the difficulty can be achieved by reducing the target contrast.
18,24
The “Macular Search Test” paradigm comprises two tasks, i.e., first to find the target, and then to identify it. Hence, this paradigm abolishes the demand for steady fixation by allowing patients to make any eye movements they need to solve these tasks. What distinguishes individual trials from each other is the location where the target first appears. The outcome measure for performance in this procedure is the response latency (i.e., the interval from target appearance to the correct response).
To allow direct comparison between patients, it was a further goal to make measuring the latency independent of target size, or visual acuity. We achieved this by letting every patient perform at a level referenced to their own acuity threshold. The results show that response latencies can be dramatically different even if visual acuity is removed as a major influence.
Preliminary reports of parts of these data were communicated elsewhere (MacKeben M, et al.
IOVS 2005;46:ARVO E-Abstract 3690).
25
One hundred thirty-five consecutive patients (44 men, 91 women) were recruited from our Low Vision Rehabilitation Service. Their age range was 18 to 98 years, with a distribution strongly skewed toward older ages (median, 80.0 ± 14 years interquartile range [IQR]). There was no meaningful correlation between visual acuity and age (
R 2 = 0.0003; see
Fig. 1).
There was a significant age difference between male and female patients (Mann-Whitney U test [MWU], P = 0.022), with the men younger (median, 77.5 years) than the women (median, 82.0 years). There was no significant sex-related difference in visual acuity (P = 0.426).
Diagnoses showed a wide range from neurologic damage after a stroke to medication toxicity, although most (97/135; 71.8%) had age-related maculopathy (ARM). This subgroup was significantly older than those with other diagnoses (MWU, P < 0.0001). Visual acuities in the better eye varied between 20/20 and 20/800 (median 20/139, i.e., approximately 0.15 visus in metric notation). There was no significant difference in visual acuity between ARM patients and those with other diagnoses (MWU, P = 0.567).
The control group consisted of 30 healthy subjects from 19 to 84 years of age (15 women) with best corrected visual acuities between 20/20 and 20/40.
The experiments were in compliance with the tenets of the Declaration of Helsinki and were approved by the local Institutional Review Board.
Subjects sat comfortably and viewed the test display from a distance of 40 cm wearing their best available optical correction. Viewing was always binocular, so that performance was mediated by the combination of the capabilities of both retinas. We deemed this acceptable, although there was the possibility that some patients might experience a small functional improvement over monocular acuity.
26,27
A beep signaled the impending appearance of a new target. The sequence of locations of appearance was randomized and unpredictable. There were 32 such locations, which were arranged (8 each) on four circles of 2°, 4°, 6°, and 8° eccentricity. Each location of appearance was used only once, so that a trial block had 32 individual trials. The subjects used a ring of 12 mm diameter (1.72°) with a center hole of 2 mm (0.29°) in the middle of the screen as a reference point by initially centering their gaze on it casually. The ring always disappeared before a target appeared.
Subjects were encouraged to use their fovea or preferred retinal locus to make any eye movements they needed to identify the target as quickly as possible. As the controlling software accepted only correct responses, recorded latencies lasted from the target appearance to the correct identification. The subject communicated the response verbally, and the examiner entered it via the computer keyboard. This indirect performance measure was deemed acceptable because it prevented contaminating the data by other variables based on interindividual differences in age, sex, and educational status.
28 –30 Instead, each recorded response latency contained a component added by the examiners' reaction times. Their medians were calculated after 200 trials for each examiner to be 633 ms (DCF) and 603 (MM). For this purpose, we used custom software that generated a randomized sequence of voice recordings of the words “up,” “down,” “left,” and “right” and required a correct response by the examiner via arrow key, as in the regular test.
In each subject, we first determined the visual acuity threshold; the program displayed a series of Landolt rings one at a time, beginning with the largest size and declining in 1/10 log steps. The patient was asked to tell the gap position for each trial. The smallest that could be identified was taken as the size threshold. This size was then doubled and used throughout the entire experiment. Thresholding typically took less than one minute.
In normal subjects, the correlation coefficient R for the sum of all latencies was 0.92, and the value for R 2 was 0.857 (85.7%). The coefficient of repeatability was 6.3% of the average.
In patients, this correlation coefficient was 0.976 and the coefficient of determination R 2 was 0.952. The coefficient of repeatability was 12.2% of the average.
It took the patients between 55 and 478 seconds to complete all 32 trials of the Macular Search test (median, 122 seconds; IQR, 82.5 seconds; i.e., typically between 2 and 3 minutes). This test duration showed no appreciable correlation with patients' age (R 2 = 0.048). Men were slightly faster (median duration, 105.5 ± 74 seconds IQR) than women (median duration, 133 ± 77.7 seconds IQR). The difference was statistically significant (MWU, P = 0.025).
The most conspicuous result of the “search and identify” paradigm was how much performance levels varied between patients. We measured this by the median latencies of correct responses, which lay between 730 ms and 10,195 ms (a factor of almost 14). The effect was equally drastic when measured by the sum of all 32 latencies per trial block, which varied between 24.4 and 404.0 seconds (a factor of >16).
The sex differences seen in the overall duration showed here too: men performed slightly better (median latency, 2410 ± 1640 seconds IQR) than women (median latency, 133 ± 1650 seconds IQR). The difference was statistically significant (MWU, P = 0.044).
In any one patient, latencies varied strongly: the longest latency could be between 1.4 and 31 times longer than the shortest, depending on where the target first appeared. There was a statistically significant performance difference between the subgroups ARM versus non-ARM with the latter being slightly faster (median latency 1730 vs. 2500 ms [MWU, P = 0.018]).
We examined the potential effect of scotomas of different sizes on search latencies. We had topographic data for 103 of 135 patients from binocular tangent screen tests. We could not use findings from scanning laser ophthalmoscopy for comparisons with data from the macular search test, because the SLO works only monocularly. Due to the limited resolution of the tangent screen test, this can only be considered an approximation. To create clear conditions, we chose 10 patients (ages 18–93 years) with no scotoma. In these cases, the target never appears in a scotoma, which should allow instant target localization and, thus, quick responses. We found the mean of the sum of all latencies to be 37.82 ± 4.86 seconds.
Conversely, we chose 10 patients (ages 50–96 years) with large scotomatous areas (e.g., through a ring scotoma, or constricted visual fields). Here, the target often appears in a scotoma, which necessitates multiple eye movements. This should cause delayed target localization and, thus, overall longer latencies. We found that their mean of the sum of all latencies was 177.30 ± 62.34 seconds.
We calculated the mean over all sums of latencies for all 135 patients as 101.92 ± 64.92 seconds. In the 10 patients with large scotomas, the mean of the sums of latencies was 177.30 ± 62.34 seconds. In those 10 with no scotoma, the same value was much shorter: 37.82 ± 4.86 seconds.
We tentatively conclude that larger scotomas are more likely to cause longer latencies than small ones. Thus, going from the average to data from patients with a large scotoma increases the latency by a factor of 1.75. Going from the average to data from patients with no scotoma decreases the latency by a factor of 0.37.
The test-retest reliability for both subject groups turned out to be very good. Regarding the coefficient of repeatability (CoR), it is no surprise that the value for the patients (CoR = 12.2%) was higher than that for the normal subjects (CoR = 8.7%), given the high variability between patients based on their fundamental differences in the topography of vision loss and different diagnoses (12 of 20 had ARM, the others glaucoma, retinitis pigmentosa, diabetic retinopathy, etc.). As the calculation of the CoR uses the SD of the mean differences of latency, which is naturally high for the patients, the value is driven up. Hence, we conclude that in these cases, the conventional pairwise correlation coefficient is the most appropriate measure of test-retest reliability, which was excellent.
The most important finding of the reported research is that search test performance varied greatly despite the fact that visual acuity was neutralized by relating the acuity demand to the individual thresholds. Furthermore, it was surprising to see that the major correlations were essentially the same for the complete cohort with a wide range of diagnoses and for the subgroup with ARM. An additional finding of interest was the fact that the age of the patients did not seem to significantly influence performance. Taken together, all three points support the notion that “search and identify” performance is influenced by factors other than age, visual acuity, and diagnosis. In connection with the points regarding test-retest reliability (see above), we conclude that the found variations between patients are truly patient characteristics and not an effect of poor test-retest reliability.
It is not surprising that some of the patients could read with acceptable speed even with a dense central scotoma in both eyes, which has also been found by others.
35 –38 Note that the current results do not allow direct conclusions from the results of detailed microperimetry by SLO, because the latter can only be performed monocularly. Because reading is a learned behavior, it cannot be ruled out that the found differences might have been influenced by interindividual differences either before onset of low vision, like educational status, or those including low vision, like current reading habits. This indicates that maintaining some reading practice, albeit with adequate magnification, may still pay off for patients with low vision.
Lott et al.
39 found that good high-contrast acuity does not assure that elderly subjects (58–102 years) can read satisfactorily and that age alone is not a good predictor of reading performance. Thus, one could have expected that our results may come out differently because of the wider age range of our patients (18–98 years) and of their universally compromised vision. However, our results showed the same (i.e., age alone was not a good predictor of reading performance). This can tentatively be explained by the fact that the presence of low vision in our cohort may have simulated the compromising conditions that led Lott and colleagues to their conclusion (i.e., low contrast vision, motor ability, and attentional field integrity).
A recent investigation relating visual capabilities and cognitive status to real-life tasks has shown that performing well in some tasks puts more emphasis on normal functioning in the cognitive domain.
40 Although we did not formally test for cognitive status here, the psychological profile taken from all patients in this study made sure that no patients with conspicuous cognitive deficits were included. Thus, it is unlikely that deficits in cognitive abilities may have emerged as a major factor.
The presented findings show that measuring visual acuity alone in patients with low vision could have led to entirely misleading conclusions. They demonstrate that there are other factors that influence performance in a task that bears resemblance with those that have to be faced daily by patients with low vision.
Eye movements were allowed in the search paradigm as well as in the reading task, but we did not monitor them in these experiments. We conclude that continuous text reading and the “search-and-identify” paradigms share an important behavioral mechanism that determines performance in both tasks. We hypothesize that the factor enabling patients to perform well in both paradigms is oculomotor control and/or eye movement strategy.
It could be argued that the high correlation of the current results with reading performance indicates that a conventional reading test can yield the same results. However, the “find-and-identify” paradigm has four distinct advantages relative to reading:
-
It is independent of the ability to read, so that very young and illiterate patients can also be tested.
-
In patients who can read, differences between levels of reading skill and habits introduce a source of noise into the data that can be avoided in the search paradigm.
-
As long as patient and examiner can communicate verbally, this method allows comparisons between patient cohorts who speak different languages.
-
It demonstrates the spatial position of where goal-directed training could intervene, so that patients could be trained to direct exploratory and compensatory movements in those directions, in which the latencies are found to be longest.
9,10
Because steady fixation is not required, the “search-and-identify” paradigm allows functionally relevant vision testing in a relaxed atmosphere and at a wide range of ages, acuities, and reading skill levels.
Supported by the Beatrice Brandes Low Vision Research Fund and The Smith-Kettlewell Eye Research Institute.
Disclosure:
M. MacKeben, MMTest (F, I, E, C);
D.C. Fletcher, None
The authors thank Alexander Gofen, for doing all the programming work for this project; Lori Lott, PhD, for assistance with the multiple regression analysis of the data; and two anonymous reviewers, for stimulating improvements of the report.