Visual acuity (VA) expresses the level of ability with respect to the spatial resolution of the visual system, and it is the first test in the analysis of the visual function.1 A number of techniques are used to measure visual acuity using different instruments, charts, and methods.2,3 Depending on the age of the subject, various optotypes (i.e., the symbols used in the assessment) are available for measuring visual acuity, each one having specific characteristics. In its evidence-based clinical practice guidelines, the American Optometric Association suggests the use of letter charts (Snellen or ETDRS) with adults and symbol charts (e.g. LEA Symbols) or small sets of letters (HOTV) with preschool children.4,5 In recording data, the letter-by-letter scoring method is preferred because it takes into account each symbol and provides a precise evaluation of VA.1,2

During vision screening aimed at detecting amblyopia, refractive errors and other conditions, it is necessary to use acuity charts with standardized optotypes and methods.6–9 The selection of the most appropriate VA charts depends on the age of the participant. Special care must be taken with toddlers and preschool children.4 The use of symbols optotypes (e.g., LEA Symbols) and small sets of letters (e.g., HOTV) are considered to be best practices for visual acuity testing of children between 3 and 6 years old.5,10 However, despite this, different optotypes and charts/cards may be used for preschool children, such as Kay Pictures,11 Cambridge crowded cards,12 Glasgow acuity charts13 and Patti-Pics.14 With school children, a full logMAR chart with different optotypes can be used.15–17

Oriented optotypes, such as Tumbling E, can be used with preschool children18 but this optotype requires left-right comprehension.19 This involves a complex mental process, especially at ages between 3 and 4 years.20 A printed card that a child can rotate, 21 or the use of a three-option forced-choice task (i.e., Up, Down, Left/Right), without the problem of left-right orientation, represent possible solutions.22

LEA Symbols are a valid alternative to the tumbling E.19 These symbols were created to improve the testability of preschool children (3–5 years old) who are not always sufficiently familiar with letters or directional symbols. They can answer by naming a picture or pointing to a symbol on a response card. The optotypes are four simple figures; square and circle and their derived apple and house. The design of the test is such that each symbol has a similar threshold for VA, and, consequently, this chart meets the criteria for a good VA test. It should be noted that the size of each symbol was slightly modified to meet these criteria.19

Digital or computerised eye charts are becoming standard in vision examinations. With digital eye charts, it is possible to modify each parameter in the presentation of optotypes such as contrast, time of presentation, presentation modalities, the spacing between targets (i.e., to control the crowding phenomena), and obviously, optotypes. Using the advantage of the random presentation of optotypes, memorization problems are abolished even for a single level of VA. With these instruments and specific procedure, simple, accurate and repeatable measurements can also be made easily with low expertise in VA measurement possessed by the examiner.1

In fact, a simple way of achieving an increased standardization of the measurement is to use digital eye charts based on algorithms that eliminate the examiner's involvement from the interpretation of the subject’s answers and the scoring of the results. It also allows a precise comparison of the results measured by different professionals.23 Furthermore, computerized VA systems can reduce the test-retest variability through repeated measurements. In this way, it is simple to detect real clinical changes in individual subjects.24 The automated scoring method does not allow the examiner to make mistakes or misunderstand the answers.25

Digital eye charts allow a QUEST function to be used, which is a psychometric procedure that defines the relationship between physical stimuli, in this case, optotypes dimensions, and the probability of a psychophysical response.26 This method identifies the most probable VA threshold value by using an adaptive process. The size of the stimulus in the next trial is determined on the basis of the subject’s answer in the previous trials.27

Different digital eye charts present different characteristics in the possible modalities of presentation, crowding, and flanker used. For these reasons, the various digital eye charts could present different results even for the same instrument. The aims of this study are to use a digital eye chart: (i) to examine the differences in VA measurement using different optotypes (i.e. LEA Symbols, Landolt Rings and Tumbling E); (ii) to compare different procedures (i.e., single presentation, line, block and QUEST); (iii) to assess the validity and repeatability of the measurement. In order to avoid response factors present in the testing of children,28 only adults were tested.

In the second experiment, the aim was to compare the different presentation modality of the digital eye chart with a classic paper eye chart and their test-retest repeatability.

Results

Descriptive results are reported in Table 2.

Table 2.

Mean ± SD values of VA obtained by different procedure of testing and session for experiment 2.

Method	Test	Retest	Difference
Single	−0,105 ± 0087	−0,120 ± 0086	−0,014 ± 006
Line	−0,108 ± 0075	−0,120 ± 0088	−0,012 ± 0042
Block	−0087 ± 0072	−0101 ± 0096	−0014 ± 0056
Quest	−0,132 ± 008	−0,134 ± 0085	−0,002 ± 0038
Paper	−0,206 ± 008	−0,210 ± 0078	−0,004 ± 0055

Firstly, a 2 × 5 repeated measure ANOVA was performed to compare all VA values. The factors applied in the ANOVA were Session with two levels (first and second), and method with five levels (single, line, block, QUEST, and paper). The results showed a significant main effect of Session (F(1,51) = 4.19, p < 0.05, η2p = 0.08), Method (F(1,51) = 72.21, p < 0.0001, η2p = 0.59) and no significant interactions were found. In order to examine the difference between methods of presentation in the best way, a repeated measure ANOVA was performed only using the first session of administration. Results were significant (F(4,204) = 66.71, p < 0.0001, η2p = 0.57). Post-hoc comparisons with Bonferroni correction show significant differences between QUEST and block (p < 0.05) and between paper and all the other four modalities (all ps<0.0001). The paper presentation shows the best values of VA. The mean difference between Line and Quest presentation was 0.024; this difference between Line and Quest presentation was significantly lower [t(102) = 6.4, p < 0.0001] compared to experiment one, but in the same direction. The results are shown in Fig. 3.

Figure 3.

Mean results for the interaction between Method × Session in experiment 2. The different points express the mean acuity values separated by method (Single, Line, Block, Quest and Paper) and Session (first and second). Bars represent +/- 1 SEM.

(0.06MB).

Regarding the repeatability, the first analysis took into account the direct mean comparisons of each modality of presentation between sessions. Even in the previous ANOVA Session was significant; direct comparison with paired t-test (without correction for multiple comparisons) showed a significant result only for line presentation (t(51) = 2.12, p < 0.05). In contrast, all other comparisons between sessions did not show significant differences. Repeatability of measurement also needs to be analysed with correlation and Bland and Altman limits of agreement (LoA). Correlation of each VA method of presentation was performed with the single score ICC [ICC(A,1)].32 The results showed a moderate to high correlation for single, block, and paper modality of presentation, but high correlation for line and QUEST (Table 3).

Table 3.

Results of repeatability of VA for all modalities of presentation. The correlation was the intraclass correlation (ICC); p-values were referred to the significance of correlation; Bias = mean difference between sessions; LoA = limits of agreement.

Method	Correlation (95 % CI)	p-value	Bias	LoA
Single	0.75 (0.6 to 0.85)	<0.0001	−0,014	0.118
Line	0.86 (0.77 to 0.92)	<0.0001	−0,012	0.082
Block	0.78 (0.64 to 0.87)	<0.0001	−0,014	0.109
Quest	0.90 (0.83 to 0.94)	<0.0001	−0,002	0.074
Paper	0.76 (0.61 to 0.85)	<0.0001	−0,004	0.107

The mean bias between sessions is smaller even than a single letter and the QUEST modality presents the smallest value. The limits of agreement (LoA) were calculated as 1.96*SD of differences. The broader results (single, block, and paper) are in the region of 5–6 letters, and the narrower (line and QUEST) were approximately 4 letters. Graphical representations of these results are shown in Fig. 4.

Figure 4.

Bland and Altman limits of agreement representation for test and retest for the five modalities of presentation in experiment 2. Values are in logMAR. The x axis represent the mean of the measurement of the two sessions. The y axis represent the second measurement minus the first measurement. The small dotted line represents the mean bias and the long dotted lines represent the positive and negative LoAs.

(0.18MB).

Digital eye charts are modern instruments that are intended to allow different types of presentations at a distance in vision examinations. However, few studies have taken into account its use and made comparison between various presentation modality and between different optotypes and have assessed the repeatability of the measurement.

Specifically, the aims of this study were to use a digital eye chart: (i) to examine the differences in VA measurement using different optotypes (i.e. LEA Symbols, Landolt Rings and Tumbling E); (ii) to compare different procedures (i.e. single presentation, line, block, and QUEST); (iii) to assess the validity and repeatability of the measurement. The study has produced some interesting findings, and these are discussed below.

Since we are interested in comparing VA between different modalities of presentation, we have used the habitual correction. Consequently, our acuity could be better defined as habitual correction visual acuity (HCVA) and not best-corrected visual acuity (BCVA), and we, therefore, cannot generalize and compare these results. The results of a study that collected BCVA values at various ages have reported values that are slightly better than ours (i.e., when referred to the single line presentation).33

A study aimed at comparing LEA Symbols with Tumbling E on a group of preschool children found comparable VA results. However, the difference between the two charts was statistically significant only in the 3–4 and not in the 5–6 year age group.20 Based on a group of adults between 17 and 31 years old, another study found that LEA Symbols showed mean VA values that were weaker than Tumbling E, with about a value of 0.042 logMAR, which is approximately two symbols.16

In a group of children of 4–9 years of age, the comparison between LEA Symbols and other pediatric tests (i.e., single line or multiple lines surrounded by a black box) with the full logMAR ETDRS chart showed that the ETDRS charts gave worse VA results compared to LEA Symbols (i.e. about 1–2 lines). Since the authors have used different eye-chart layouts, the crowding effect had a crucial role and interacted with different layouts and optotypes.29

Different modalities of crowding affect the VA thresholds. Sailoganathan et al.3 found different results for LEA Symbols surrounded by flanking optotypes or flanking bars.

The crowding phenomenon is an important factor in measuring VA because of the different spacing between optotypes produces different VA values.15 Visual acuity charts, cards, and tests offer several modalities that can be used to control the crowding effect by means of different flankers and spacing.28 As a result, data from different charts are not directly comparable and this should be taken into account when comparing charts.3,16,29

In the first experiment, when comparing optotypes, we found that the LEA symbols show a worse VA of about 1 line. We also found that the QUEST presentation for Landolt Rings and Tumbling E optotypes show better values of VA compared to the single-line presentation using a letter by letter scoring. This result is also confirmed in the second experiment, but with a smaller difference. Comparing single letter, line and block presentation directly, we found that only block presentation shows lower values of VA. Crowding has a great influence when a target is flanked in four directions rather than two.

The unexpected result is the significant difference (0.07 logMAR) between the other formats and the paper charts in the second experiment. The reason that a paper eye-chart shows better results is an interesting question. All instruments were correctly set at the correct distance from the examiner. Contrast does not seem an acceptable explanation because it is greater for the digital optotype than the paper chart. The most realistic explanation seems to be linked to the different distances used for testing. The VA at 5 m (and not at 3 m), can be influenced by a small degree of myopia / pseudo-myopia, and this factor could explain the difference. In fact, one of the limitations of this study is the testing of HCVA and as opposed to BCVA, and consequently, the role of the refractive status is unknown.

Digital eye charts based on different algorithms have been compared with standard charts. Harris et al.25 did not find any significant difference when comparing backlit ETDRS with an automated ETDRS system (A-ETDRS). The A-ETDRS measures VA in two phases: range finding and thresholding. The same two phases are used by the compLOG system. No significant bias was found between the gold standard ETDRS and the compLOG system, the time required for testing being similar.34 The compLOG was also used with Kay Pictures and to compare the printed Kay Pictures chart with compLOG Kay Pictures, resulting in the detection of a small bias. However, the difference was small and not clinically significant.35

The agreement for all modalities of presentation of VA was excellent, with a small mean improvement over time (with a value of half of a symbol). This bias between sessions is aligned with a study that reports little bias 36 but smaller than the previously reported data of about one line.37,38 Compared to previous studies,36–38 which report LoAs of about 0.15−0.18, the LoAs found were smaller (0.07−0.12) with the smallest values for QUEST and line presentation. The high correlation and small LoA is a sign of consistent and stable performance over time. The correlation value has an essential role for its connection to the standard error of measurement, test-retest variability, and, consequently, the smallest visual acuity changes that can be detected.39 However, on comparing the different methods, QUEST shows better results in terms of smaller LoA and bias. Based on the overall results of experiment 2, excluding QUEST, the single line seems to be the better modality of presentation, bearing in mind a difference between this and block presentation of about one letter.

From a psychophysical point of view, the QUEST presentation represents the best modality of letter presentation, preventing the response factor and the influence of the experimenter. However, the line presentation presents an advantage because by using the remote control of the digital eye chart, it is simple to switch from the single letter presentation to reach the threshold zone quickly, and then switch to line presentation in order to measure the threshold precisely using letter-by-letter scoring. This is perfectly consistent with the presentation method used in the A-ETDRS,25 and the FrACT systems.40

The limitations found in the digital optotype used are in the totally un-flanked presentation during QUEST and the lack of line of -0.4 logMAR in the block presentation. We hope that future revision of the software could include the QUEST presentation with flankers (using the same optotype) and another line of optotypes in the block presentation. A limitation of our evaluation is the use of only adult participants. This point could be solved in future studies with children participants to provide clinical information specific for this population.

In evaluating VA, digital eye charts present different modalities of presentation and optotypes. Clinicians who use digital eye charts have to take into account that not all modalities and optotypes are equivalent and give different VA thresholds. Landolt Rings and Tumbling E present similar thresholds while those of LEA Symbols are worse. The modality of presentation also affects the measurement of VA, which was stable over time. Taken together, these aspects of the measurements become essential in patients that present VA near the cut-off when specific thresholds are required (e.g., driver licensing, specific jobs, etc.). Under these conditions, the differences in the type of presentation and the optotypes used play an important role.

The authors have no conflicts of interest to declare.