- Top of page
- MATERIAL AND METHODS
Introduction: The most common way to examine vision is to measure visual acuity using letter charts. The Rarebit fovea test was developed for detection of small defects of foveal function at a stage before they cause abnormal visual acuity. In a recent study, the RFT was well tolerated in a smaller group of children between seven and nine years of age; however, the number of subjects in that study was small for the determination of reference values, and therefore the aim of the present study was to establish reference values for a larger group of children aged between six and 10 years and to evaluate the learning effect after repeated tests.
Methods: Rarebit fovea test data were collected from a group of 108 subjects aged between six and 10 years as part of a screening program at a compulsory school in Stockholm, Sweden. All subjects had good visual acuity and minor refractive errors. They underwent two Rarebit fovea test examinations on one occasion.
Results: Rarebit fovea test results ranged from a median mean hit rate of 94.0 to 97.5 per cent for pre-school and third-year children, respectively. For the entire group of children the median mean hit rate was 96 per cent (range 57 to 100). The third-year children performed significantly better than the pre-school (p < 0.01) and first-year children (p < 0.05). A significant improvement from the first to second test run was noticed in all groups.
Conclusion: The Rarebit fovea test proved to be well tolerated among children in a group of six to 10 year olds and more than 90 per cent of children were considered to give reliable results close to what is normal for adults. It would be interesting to further investigate the potential of the Rarebit fovea test for evaluation of foveal function in children.
The most common way to examine vision is to measure visual acuity (VA) using letter charts. Even in quite young children, letter or symbol VA is used as a measure of visual ability.1 As soon as a child is able to read, normal letter charts like the standardised Early Treatment Diabetic Retinopathy Study (ETDRS) chart can be used and expected monocular decimal VA is 0.8 to 0.9 in children aged six to eight years, and 1.0 in those aged eight to 10 years;2,3 however, VA testing using conventional optotypes is insensitive for the detection of subtle defects in visual function. Conventional letter charts present targets that are large compared with foveal receptive fields4 and a decimal VA of 1.0 can be sustained with less than two-thirds of the normal number of optic nerve axons.5
Rarebit perimetry (RBP) is a relatively new computerised technique that relies on very small stimuli.6 Contrary to the conventional threshold perimetric approach, the RBP aims to detect the presence of function rather than estimating the level of function. The test is sensitive to small gaps in the receptor matrix, because the stimuli used are small in relation to the receptive fields, unlike conventional stimuli such as the Goldmann size III. Another way of describing the nature of the test is that RBP ‘probes the integrity of the neural matrix’.6 Hackett and Andersson7 recently evaluated this principle by examining healthy subjects and subjects with glaucoma using RBP with different luminance settings. They found that elements (that is, photoreceptors communicating through the same ganglion cell and its cortical channel), which could be assumed to be dysfunctional, were able to respond to the very intense stimuli of RBP and in a way contradicted the notion that the percentage of stimuli seen is an approximation of the proportion of normal elements remaining. In contrast, their glaucoma group showed poorer test results at each stimulus luminance level compared with the healthy subjects, although the difference between the groups was more pronounced using reduced stimulus luminance.7
This new approach for detecting subtle defects of the visual system has been found suitable for neurological diagnoses and glaucoma.6,8,9 Furthermore, RBP can detect reduced visual function in prematurely born children with damage to the white matter of the brain due to immaturity,10 despite having normal VA.
The Rarebit fovea test (RFT) is included in the Rarebit perimetric package and was developed for the detection of small defects of foveal function.6 When starting the Rarebit perimetric package, the examiner initially has to choose between perimetry of the central field (30°), perimetry of the right or left flank (60°) or foveal testing (the central four degrees, that is, the RFT). The RFT has been evaluated less than the RBP but is useful for the detection of small defects of central visual function. Agervi, Nilsson and Martin11 detected subnormal foveal function in the fellow eyes of children treated for amblyopia. In line with these results, other studies have found decreased visual function in the non-treated eyes in amblyopes using contrast sensitivity tests,12 visual evoked potentials13 and microperimetry.14 In adults, the RFT reveals foveal dysfunction, which correlates with structural changes detected with optical coherence tomography in diabetic patients without obvious vascular abnormalities.15
In addition to the small stimuli (less than 0.5 minutes of arc), a special characteristic of the test is its use of intense stimuli that are briefly presented against a dark background.
The test time is short, only one to two minutes and the area tested is the most central visual field, four degrees horizontally and three degrees vertically. The results are given in percentage as ‘mean hit rate’ (MHR), which is equivalent to the number of dots seen in relation to the number of dots presented. Reliability is estimated by counting false-positive answers, which are given as the number of errors. Ten per cent of the stimuli presentations are used for this purpose.
Uncomplicated test principles and short test time are beneficial qualities when examining children, because psychophysical testing is often associated with difficulties in getting the child to participate with sustained concentration.16 Furthermore, threshold testing is a demanding task, while supra-threshold stimuli, like those used in the RFT, improve compliance.6,17
The RFT resembles a simple computer game and the test principle has been found somewhat amusing by children,18 something that could help to prevent lack of attention and fatigue and thereby improve co-operation and reliability.
Agervi, Nilsson and Martin11 used the RFT to evaluate foveal function in children treated for amblyopia and found that the test was well tolerated (that is, easily understood and interesting, thereby resulting in good compliance and reliable results) by children between seven and nine years of age. In that study, the median MHR in the control group was 97 per cent (range 87 to 100) and the median number of errors was one (range zero to four). In healthy adults between 22 and 65 years of age, the expected median MHR has been found to be 100 per cent (range 97 to 100)19 and the median number of errors was zero (range zero to two).
During the first years in school, visual demand increases as letter and line spacing decrease.20 Therefore, it is crucial to detect abnormalities in visual function by the use of reliable, cost-efficient tests, which are well accepted by children. Conventional VA tests would be the first choice for visual function screening but if perceptual or other visual problems are present, complementary tests should be used. The RFT has the potential to serve as such a test. The number of subjects in the study by Agervi, Nilsson and Martin11 was somewhat small (n = 25) for determining reference values, and therefore the aim of the present study was to establish reference values for children aged six to 10 years and to evaluate the learning effect after repeated tests.
MATERIAL AND METHODS
- Top of page
- MATERIAL AND METHODS
A group of 181 subjects aged six to 10 years were examined in a screening program at a compulsory school, The Eira School, in Stockholm, Sweden. Informed consent was obtained from the parents and the study adhered to the Declaration of Helsinki. Children from pre-school to third year were examined and the screening program included cover test, colour vision testing, auto-refraction and VA. The auto-refractor used was a Shin Nippon NVision-K 5001 (Shin-Nippon Commerce Inc, Tokyo, Japan). The system allows binocular open field distance fixation, which is thought to eliminate potential instrument-induced myopia. After auto-refraction monocular VA was measured with an ETDRS test chart at a distance of four metres and the result was noted as the number of letters read and decimal VA.3,21
Because providing appropriate refractive correction was not within the framework of the screening program, subjects with refractive errors that could influence the RFT results were excluded.
The criteria for participation in the present study were: decimal VA of at least 1.0 in one eye; refraction within +1.0 and -0.5 dioptres; astigmatism up to -0.5 D; no amblyopia, strabismus or colour vision defects. Of those examined, 108 children fulfilled these criteria and were examined with the RFT. The included subjects were divided into four subgroups according to what school year they were in. Swedish children normally start pre-school the year they turn six years of age, that is, age and school year are closely related. Demographic data are given in Table 1.
Table 1. The number of subjects participating in the study and their median age given in months and visual acuity (VA) given as number of letters read on an Early Treatment Diabetic Retinopathy Study chart
| ||Pre-school||1st year||2nd year||3rd year|
|Number of subjects||25||29||24||30|
Rarebit fovea test
The RFT (version 4) was performed using a personal computer and a 17-inch liquid crystal display in a completely dark room at a test distance of two metres. The test briefly presented two intense (luminance 158 cd/m2) small dots (less than 0.5 minutes of arc) separated by one degree of arc against a dark background (less than 1.0 cd/m2). In the present study, the screen was calibrated before each session. One standard examination consisted of five test runs, in which a total of 100 stimuli were presented. The contrast was kept constant and the subject was instructed to fixate on a flickering cross in the centre of the screen. Each presentation was preceded by a sound to enhance attention. According to the manual, 10 per cent of the presentations are used for controls, in which one or no dots are presented and the program automatically increases the number of control presentations for subjects who tend to give false-positive answers. To evaluate this automatic function, a pilot study was conducted. In the pilot study, a total of 20 examinations were performed, each consisting of five runs. During the first 10 examinations, in which the subject (one of the authors) gave all correct answers, the number of control presentations ranged between three and 10, with a median of 4.5. During the next 10 examinations, false-positive answers were given at every control presentation. During this session, the number of control presentations ranged between four and eight, with a median of 5.5.
The children included were carefully instructed to indicate the number of dots seen (none, one or two) by clicking on a mouse button (no click, single click or double click). All participants underwent two examinations to ensure that the subjects fully understood the task. Only one eye was tested, that is, the eye with better VA or smaller refractive error.
The examiner, one of the authors (MS), observed the subject during the whole examination and when necessary encouraged the subject to maintain attention. No special rewards were offered and the program was not adjusted in any manner to better suit children. The examiner also subjectively evaluated and noted both the subject's ability to give a double click and their tendency to be trigger happy, that is, click before the stimuli were presented. Subjects with either type of clicking problem were classified as having ‘questionable performance’.
The difference in MHR between the first and second examination was analysed with a Wilcoxon matched pair test. The RFT data should be analysed with non-parametric tests, because a ceiling effect is expected, which gives data that is not normally distributed. Comparisons of the MHR among different age groups were analysed using non-parametric repeated ANOVA. For correlations between MHR and age and MHR and VA, the Spearman rank correlation was used. To define reference values, median MHR and lower confidence intervals were calculated. The Chi-squared test for independence was used when evaluating incidences of ‘questionable performance’. A p-value of 0.05 or less was regarded as significant. Based on previous results with the RFT, it was determined that a minimum of 21 subjects in each age group was needed for 90 per cent power based on Lehr's formula.22
- Top of page
- MATERIAL AND METHODS
For the purpose of establishing reference values, data from the second test (Test 2) were used to minimise the bias that test unfamiliarity might have had on the MHR. The derived reference values ranged from a median MHR of 94.0 per cent in the pre-school group to 97.5 per cent in the third-year group (Table 2). For the entire group of children, the median MHR was 96 per cent (range 57 to 100). Lower confidence intervals were calculated to define reference data and are presented in Table 2. Over ninety per cent of the children had a MHR better than 85 per cent and more than 82 per cent of the children had a MHR better than 90 per cent (Figure 1).
Table 2. Results from examination with the Rarebit fovea test (RFT). Data are presented in median and range values. The RFT result is presented as mean hit rate (MHR) given as a percentage. Lower 95% CI = lower 95% confidence interval for MHR given in percentage. Errors are given as total number of false positives and ‘test time’ as the total testing time in seconds.
| ||Pre-school||1st year||2nd year||3rd year|
|RFT||Test 1||Test 2||Test 1||Test 2||Test 1||Test 2||Test 1||Test 2|
|Lower 95% CI|| ||88|| ||87|| ||93|| ||92|
|Test time (s)||115||105||107||102||105||100||103||99|
Figure 1. Categorised mean hit rate results for all subjects. The bars illustrate the number of subjects within the four different mean hit rate categories. Eighty-two per cent of subjects achieved a mean hit rate of 90 per cent or more and 61 per cent achieved a mean hit rate of 95 per cent or more.
Download figure to PowerPoint
The third-year children performed significantly better than the pre-school (p < 0.01) and first-year (p < 0.05) children (non-parametric ANOVA, Kruskal–Wallis test).
An improvement from the first to the second test was noticed in all groups, pre-school (p = 0.0001), first year (p = 0.0056), second year (p = 0.0002) and third year (p = 0.01) (Wilcoxon matched-pairs signed-ranks test).
Median MHR did not differ significantly between girls and boys (p = 0.7) (Mann–Whitney).
No statistically significant differences were found in the numbers of errors or the total test time among the four age groups. The number of subjects with ‘questionable performance’ decreased (p = 0.0024, Chi-squared test for independence) with increasing age (Table 2).
There was a weak but significant correlation between MHR and age (p < 0.0001; r2= 0.17, Spearman rank correlation) (Figure 2) and an even weaker correlation between MHR and VA (p = 0.0093; r2= 0.063) (Spearman rank correlation).
Figure 2. Correlation between age (in months) and mean hit rate. There was a weak but significant correlation between these parameters (p < 0.0001; r2= 0.17).
Download figure to PowerPoint
- Top of page
- MATERIAL AND METHODS
The evaluation of visual function is important when children start school, because visual demand increases with increasing age. A new test for the evaluation of foveal function, like the RFT, could provide new information and offer extended possibilities for the follow up and detection of subtle defects of the visual system in, for example, prematurely born children with perceptual disabilities or in the fellow eye of children treated for amblyopia. These are conditions in which careful evaluation of the visual system and foveal function might give important information to help characterise the nature of a disease. The results from the present study indicate that the RFT gives reliable results when examining normal children.
The median MHR found in this study (96 per cent, range 57 to 100 per cent) is somewhat lower than the median MHR established for adults (22 to 65 years old),19 but is in line with the results from Agervi, Nilsson and Martin.11 The difference between children and adults and between children in different age groups could depend on the maturity level of the visual function. By using a contour-detection task to evaluate spatial integration, Kovács and colleagues23 found significant improvement in the visual function of children aged between five and 14 years. They explained the improvement in terms of a protracted development of ventral visual-stream function in humans. It is difficult to judge whether that is a reasonable explanation for our findings or whether they are simply due to children's reduced ability to maintain attention compared with adults; however, an improvement of performance during psychophysical tests depending on age is well known.3,24,25
A weak correlation between MHR and VA was found but has not been identified in other studies.11,19,26 This correlation was extremely weak and is most likely to be explained by the fact that both VA and MHR increased with age.
The third-year children (nine to 10 years of age) performed significantly better than the pre-school children (five to six years of age) and first-year children (between six to seven years of age). Seven out of 25 children in the pre-school group were classified as having questionable performance compared with only two, one and zero children in first, second and third year, respectively. Other studies have shown children from five to six years of age being able to perform remarkably well during VA and perimetric testing; however, performance improved with age and children from eight years were found to perform even better.3,16,24 Our results are in line with these data.
An improvement in MHR between the first and second examinations was found in all subject groups, which might be explained by the lack of a familiarisation procedure. The children in the present study were only given verbal instructions before the RFT began. In other studies, familiarisation procedures have turned out to be crucial for performance and suggested that they should be mandatory in children younger than seven years.24 Therefore, a demonstration of the RFT, as recommended in the test manual, should always be applied before the examination.
Regarding reliability, the number of false positives (errors) did not differ between groups. According to the test manual, 10 per cent of presentations are used for control purposes. Nevertheless, during the pilot study (Method section) it was discovered that the number of control presentations varied between six and 20 per cent throughout a series of 20 test rounds. Therefore, it might be difficult to draw any specific conclusions regarding the reliability between the groups. The median number of errors in the entire group was two compared with one found in children between seven and nine years11 and zero in a group of adults19
An increased number of control presentations could be applied to enhance the reliability of the RFT. Safran and colleagues24 used numerous control presentations to avoid a non-selective automatic answer pattern. In contrast, an increase in control presentations will make the test more time consuming and boring, which in turn might cause fatigue and reduced reliability. We would like to suggest using the same number of control presentations, that is, that the number of control presentations do not vary during each test round when testing children. We sensed that when only a few control presentations were applied, subjects occasionally lost concentration and became ‘trigger happy’; however, when testing children, the examiner's subjective assessment of the child's performance might provide additional clues as to whether the results are reliable.
Regarding the technical device, a mouse button might be less suitable for the youngest children, because in a few cases their small hands and short fingers made it difficult for them to double click. Replacing the mouse with a simple button or a game control would most likely eliminate this problem.
One could argue that the large number of excluded subjects, mainly due to small refractive errors, might bias the results; however, studies27,28 have reported on the negative effect of optical defocus on perimetric results. Small stimuli, close to fixation, like the ones used with the RFT, will result in a reduced MHR with increased standard deviation in the presence of optical defocus.28 Subjects with VA less than 1.0 were also excluded, which might have caused a selection of children with ‘super-normal vision’. In contrast, the children in the present study showed a better VA in general compared with values reported for children in the same age group.3 Because there was a low level of correlation between MHR and VA in this study, the bias effect from having tested subjects with ‘super-normal vision’ is likely to be trivial.29 Therefore, it is unlikely that this has biased the results in such a way that they cannot serve as reference values for the RFT.
The lower confidence intervals were similar in all groups, ranging between 87 and 93 per cent. Therefore, a MHR of 90 per cent or more is a reasonable approximation of the 90 per cent confidence limit across age groups. Using each group's individual limit would provide more accurate results when examining foveal function in the age group of six to10 years. The test was designed to reveal subtle defects of foveal function and therefore we suggest that the median MHR for specific ages should be used as normal/abnormal cut-off values if other signs of abnormal visual function and/or structure have been found.
The principles for the MHR are uncomplicated and easy to understand, even for a child; however, judging from the test versus retest results, a test run is crucial. Furthermore, we believe that the RFT should not be used as a ‘stand alone’ test for screening of visual function in children, but rather as an indication for the need of further investigations and/or follow up. Even if these data support the use of the RFT in the examination of children's foveal function, the specificity and sensitivity of the test in the detection of decreased foveal function among children must be evaluated to fully understand the potential of the RFT in clinical settings.
Based on previous results,11,15 the RFT should be considered as a technique that has the potential to give important information regarding foveal function in conditions where more advanced electrophysiological examinations and more demanding microperimetric tests would be of interest. The RFT could easily be incorporated in any ophthalmological or optometric setting in contrast with more expensive techniques like electrophysiology or scanning laser ophthalmoscopy (an example of an established microperimetric test). The main interest of the present study was to evaluate the acceptance of the RFT among children in different age groups and to determine normative data for these groups. We believe that the RFT could be used as an indicator of the need for further investigation when VA is normal but there are subjective symptoms or other visible structural changes detected ophthalmoscopically.