Assessment of near visual acuity in 0–13 year olds with normal and low vision: a systematic review

Study eligibility based on inspection of titles and abstracts was performed by the two authors (BH and FNB), using the inclusion criteria presented in Table 2. All stages of study selection, data extraction, and quality assessment were also performed by these two independent reviewers (BH and FNB). Disagreements during selection were solved by application of inclusion criteria, and reaching consensus after discussion.

Included quantitative studies had to adhere to the following criteria: 1) report near visual acuity scores in one of our two target groups, 2) (lowest inclusion) age of children should be between 0–13 years, and 3) the study has a cross sectional or observational design. In order to increase data collection we also searched for articles by inspecting reference lists of the included studies. Screening studies were included, but studies only involving children with amblyopia were excluded from this review since the focus is explicitly on children with normal vision and children with low vision. Children with refractive errors were included, since they can attain normal visual acuity when corrected properly.

Quality of the included studies was evaluated independently by two reviewers (BH and FNB) using criteria for cross sectional and case-control studies. Information for evaluation of the included studies was: number of participants, method used, a clear outcome definition (in this case near visual acuity scores), and results (reporting confidence intervals and thresholds in case they were presented). The first author extracted the data and contacted authors of identified studies for additional data if necessary.

Because of the scarcity of studies providing quantitative data about near vision in children and the wide range of vision tests used to measure near vision we report results in a narrative rather than a quantitative manner.

Our search in the three electronic databases provided a total of 602 citations. After removal of duplicates there were 436 studies left. After screening these studies on their titles and abstracts the number of studies for inclusion was 89. The next step was to screen full-text versions of these articles for eligibility. Fifty-five studies were excluded because they did not contain the primary outcome measure (n = 40), they were not published as an article (n = 6), or because they included the wrong population (n = 9). After doing this, 34 studies remained. Of the included studies, 29 were quantitative studies, and 5 were qualitative studies. See PRISMA flow chart Fig. 1.

Thirty-four studies were included. For the 0–3 year old group, 2 qualitative [9, 10], and 10 quantitative studies were included [11‐20]. Of these 12 studies, only one study reported near visual acuities in children with VI [15]. For the 4–7 year old group, 12 studies were found in which near vision was measured in children with NV, and 3 studies were found for children with VI. For the 8–13 year old group, 10 studies were included which measured near vision in children with NV, and 2 studies which reported near vision in children with VI.

Table 3 presents the design, participant characteristics, method and outcome of the included studies. Figure 2 provides visual acuity estimates collected with different procedures. The first main message that can be extracted from Fig. 2 is that the 70 and 75% FPL thresholds seem to be too high for young children, resulting in an underestimation of their perceptual abilities [12, 14]. A recent study also provided evidence that the optimal threshold for young children is often below 50% correct, because of high lapse rates and relatively lower rates of guessing correctly in infants [20]. Another important aspect of near vision assessment in young children (aged 2–6 years) is that optotype tests are more effective test in detecting uncorrected refractive errors than preferential looking paradigms. A disadvantage of the C-test is that it cannot always be used successfully in young children (testability is 87% in children aged 26–72 months) [18].

The golden standard for testing near visual acuity in 0–36 month old children is the TAC, providing norm scores across the whole age range and validated for children with NV [17] and low vision [15]. For the TAC, a left/right 2 alternative forced choice procedure is used with a 50% guess rate. During TAC assessments, the observer reports whether the infant directed his/her gaze towards the side with a grating pattern. A critique on the TAC is that norm scores have lower boundaries that are not representative for normal vision; they should be higher) [17]. These low boundaries affect the test’s sensitivity to pick up near vision problems. The second critique is that the visual acuity that is measured relies heavily upon the subjective judgement of an observer [19].

More than 30 years after Teller’s review, there are now developments towards a computerized version of the TAC using eye-trackers to measure gaze direction. This new method offers a more objective and standardized way of measuring visual acuity in children than the TAC, because TAC acuities rely solely on a subjective report of whether or not the child’s gaze was directed at a target. A computerized version of the TAC has several advantages: (i) it is fully automated and scoring does not rely on an observer, and (ii) it has control over key parameters that, ideally, should be standardized (e.g. luminance, presentation distance, and location of the stimulus in the visual field). However, it also has disadvantages, because certain eyes are not easy to track, for instance because of nystagmus or iris transillumination. So, a computerized application of the TAC might be very useful, but since eye-tracking is not always an option, it is likely that TAC cards will still be used in the future.

Children aged 3 years and older can reply in a verbal manner and can grasp the concept of matching an answer to a stimulus. The response manner in the majority of studies is verbal naming or matching. Thirty years ago, a review on vision testing in 3–5 year old children reported that matching works reliably in a number of studies, while young children may refuse to give an adequate verbal response to testing [6]. The number of successful measurements was not reported in the majority of included studies, so no recommendations can be made with regards to the preferable response manner for children aged 4–7 years. In the nine quantitative studies that were included verbal response manners were used.

As can be seen in Fig. 3, there is a good correspondence between the Bailey-Lovie chart used to measure near visual acuity in 3–5 year olds and the Landolt C crowded chart in children with an average age of 5 years with NV. Another observation that can be made when looking at Fig. 3 is that Landolt C acuity and LEA acuities are in good agreement with each other. The near vision chart used by Dowdeswell was the same as the conventional chart used by Huang, but in the Dowdeswell study near visual acuity appears to be better. An explanation might be that the mean age of children in the latter study was higher.

In children aged 4–7 years, near visual acuity can be measured with crowded and uncrowded acuity charts. In three of the included papers uncrowded and crowded visual acuities were compared [21, 27, 28]. In children aged 4 to 7 years crowding, i.e. a deterioration of object recognition due to nearby flankers, is a normal phenomenon. When interpreting differences between uncrowded and crowded acuity scores, one should be aware that crowding decreases with age in children with NV. In children with NV, the difference between uncrowded and crowded visual acuity is 1–2 logMAR lines on the visual acuity chart (this entails crowding ratios between 1.25 and 1.6) [22, 27, 28]. In children with NV aged 6 years and older differences between uncrowded and crowded visual acuity of ≥ 3 logMAR lines are considered to be increased [22]. In children aged 10 or older crowding is practically absent, therefore a difference of 2 or more logMAR lines between crowded and uncrowded acuity is considered to be too large [33], because in adults with NV and older children ratios of 1.2 have been reported [34].

As can be seen in Table 4, the distance at which near visual acuity is assessed, differs considerably across studies. Data from the Boonstra study show that viewing distance does affect near visual acuity measures (Fig. 4). When children look at the chart at a self-chosen distance (which ranged from 5 to 20 cm), the acuity that is measured is ~0.15 logMAR lower than the acuity measured at 40 cm. The rationale behind the 40 cm rule is that accommodation is not expected to play a large role while at shorter distances it does [4]. Furthermore, it is difficult for the experimental leader to monitor viewing distance accurately if a child adopts a self-chosen viewing distance. In general, poorer acuity for near than distance vision has been reported before in subjects with NV and was attributed to errors in accommodation, for example accommodation lags (under-accommodation) or accommodation leads (over-accommodation) [35]. Near visual acuity (Snellen equivalent) is calculated by dividing the measurement distance in meters by the M-value. If this distance is reduced without necessity, the outcome is very likely to be lower and not representative for NVA.

In children aged 4–7 years there is not such a clear golden standard as there is in 0–3 year olds. Near visual acuity can be assessed with several validated tests with comparable outcomes, for example the crowded Landolt C-test [31], the crowded LEA-version of the Landolt C-test [27] and the Bailey-Lovie chart [26]. Of these tests, the LEA-symbol is the preferred optotype, because children know the symbols, left-right confusion cannot affect measurements, symbols have been validated properly against Snellen E’s and Landolt C’s for size (LEA symbols have to be 1.5× larger than the E in order to result in the same VA), symbols have equal discriminability, and good test-retest reliability [36]. The appropriate response method is to let the child match the correct symbol or to ask the child to provide a verbal answer. Clinicians should keep in mind that viewing distance should be fixed, since small shifts in viewing distance can have a large impact on acuity, i.e. at 40 cm a 10 cm shift can cause a 25% change in visual acuity measures. Furthermore, inaccurate accommodation at smaller viewing distances is likely to exert a negative influence on the measured acuity. Finally, (near) visual acuity in children aged 4–7 years is highly affected by the presence of distractors in both children with NV and even more so in children with VI.

In the 70’s and 80’s the Sheridan Gardener test was often used to measure near visual acuity in school aged children and the Snellen chart was used to measure distance visual acuity [32, 43]. Nowadays, the ETDRS chart [38, 39, 42, 44], the Bailey-Lovie letter chart [45] and LEA charts [33, 40] are more frequently used to assess near visual acuity in school-aged children. The ETDRS and Bailey-Lovie letter chart can be used in children that can read letters. As can be seen in Fig. 5, the method that is used to measure near visual acuity has a profound effect on the outcome. The two Larsson studies used LEA symbols to measure acuity in 10 year olds and report much better acuities than the other four studies in which letters were used to measure acuity. One study compared distance visual acuity for letters with distance visual acuity for LEA symbols and found a difference between these two methods of almost 1 logMAR line [40]. Viewing distance was 40 cm for the majority of studies (6/8 studies where viewing distance was reported). In two studies a viewing distance of 25 cm was used [43, 45].

Two studies were included on near visual acuity assessments in subjects with VI [39, 45]. The goal of the Wolffsohn study was to develop a reading chart that could be used to assess near vision in a quick and accurate manner in low vision patients [45]. The authors emphasize that near visual acuity measurement is an important clinical measurement as many visual demanding tasks are performed close to the eyes, especially by patients with low vision. In addition, certain ocular disorders can have a differential effect on near and distance visual acuity. The drawbacks of other near vision charts that are mentioned by the authors are that reading charts are often slow to establish a near acuity threshold (the example being mentioned is the Bailey-Lovie chart), especially in those with reduced vision, poor language or weaker cognitive skills, and that they often lack realism to commonly performed tasks such as reading the paper or telephone book. A Practical Near Acuity Chart (PNAC) was developed to measure near visual acuity as accurately as the available charts in patients with VI, but in less time. The PNAC incorporates 8 important design features: large (N80 or 1.6 logMAR at the prescribed 25 cm reading distance) to small print size (N5 or 0.4 logMAR at 25 cm reading distance), regular progressions between lines (0.1 logMAR), an equal number of words on each row, one three-letter word one four-letter word and one five-letter word per row, the use of word sequences that are related, easy to recognize for 9-year old children, Times New Roman font (common font), and paragraphs of the most commonly used print sizes on the backside to determine reading speed and fluency. The PNAC near acuity thresholds and Bailey-Lovie near acuity thresholds did not differ, while test duration was longer for the Bailey-Lovie chart (32 ± 2 s versus 76 ± 4 s).

The second study in which near visual acuity was measured in children and adults with VI due to infantile nystagmus aimed to assess differences in near and distance visual acuity [39]. This was done because there was controversy in the literature about the effect of viewing distance on visual acuity in individuals with infantile nystagmus. It was hypothesized that, in patients with infantile nystagmus, near visual acuity could be better because of a dampening of the nystagmus intensity at near compared to distance viewing. This dampening of nystagmus was indeed observed, but there were no consistent relations between the dampening of nystagmus and an increase in acuity suggesting that visual acuity in subjects with infantile nystagmus appears to be limited more by sensory than oculomotor deficits.

The ETDRS was the most frequently used chart to measure near vision in children aged 8 years or older. There are clinically relevant differences in visual acuity measured with letters and visual acuity measured with symbols (differences appear to be ~0.1–0.2 logMAR). Therefore, clinicians should report the chart that was used to measure acuity. The relatively lower acuity for ETDRS letter charts might be caused by a relatively low guess rate (10%), and the use of SLOAN letters, which are more difficult to discriminate than resolution or symbol optotypes. The benefit of using letters instead of symbols is that letter recognition lies closer to reading ability than symbol recognition.

For 0–36 month old children, the TAC procedure is the clear golden standard. During the development of the TAC in the early ‘80s, there were studies that compared outcomes collected with different thresholds (for example 60% and 75% [12], or 55% and 70% [14]). In psychophysics, the threshold is often defined as the ‘halfway up point’ that lies halfway between the guess rate (γ) and 100% correct rate [46]. So, when applying this psychophysics threshold rule to the TAC, which in essence is a 2 alternative forced choice task (2AFC), a 75% correct threshold should be used. However, the 70% threshold collected with forced preferential looking procedure in 3-month olds [14] lies well below the lower border of the norms collected by Salomao et al. [17]. The same trend was observed when comparing the 75% FPL thresholds collected by Mayer et al. [12] with the lower bounds collected by Salomao et al. [17]. Salomao et al. used a modified staircase procedure to measure grating acuity with the TAC, which entails that cards were presented from lower to higher spatial frequencies in one octave steps up to the threshold region and then 0.5 octave steps around the threshold. Testing continued until two consecutive staircase reversals occurred. The VA threshold was defined as the spatial frequency that received two positive responses. If this could not be done in six trials (counting from the first negative response), threshold was defined as the highest spatial frequency that received the greatest number of positive responses. In the 4–7 year olds, only 4 studies reported the scoring procedure that was used. The Hohmann study in which the C-test was validated used an 88–94% threshold, the Huurneman studies used a 60% threshold, and Lovie-Kitchin scored the acuity per letter (5 letters per line 0.02 logMAR per letter on a row). Studies in the oldest age group did not report which thresholds were used. Considering the large differences in outcome between studies, threshold choice might have affected vision outcome; using a high threshold (75% or higher) can result in lower acuities than using a lower threshold (50–60%). Standardization is needed to enable comparison between different acuity scores. For a 4AFC task, such as the C-test or the LEA symbols, a 62.5% threshold lying halfway between the guess rate and 100% correct would be preferable [46].

For children aged 3 years and older it is preferable to use charts that are more sensitive than the TAC [18]. The Landolt C test is more sensitive in detecting uncorrected refractive errors than the TAC and results in significantly lower acuity estimates [18]. The explanations that the authors offer for the lower Landolt C versus TAC-scores are: task complexity, acuity gradation (inter-card intervals TAC correspond to 2–3 acuity lines of the C-test), and the higher sensitivity of the C-test to detect uncorrected refractive errors. In addition, differences in testing procedure might also affect measurements: the uncrowded C-test offers 6 optotypes per row and the authors allowed one mistake per row [18]. This means that thresholds for the C-test were higher than for the TAC test, which could in itself result in poorer acuity estimates measured with the C-test compared to the TAC acuity. In addition, the TAC and Landolt C test are both examples of tests measuring resolution acuity, but the Landolt C ring has higher sensitivity for detecting uncorrected refractive errors than the TAC and in case of the Landolt C matching is possible which provides a more reliable response. Explanations for this could be that the TAC stimulus is larger (12.5 × 12.5 cm) than the C-test stimulus. Lower acuity estimates collected with the C-test than the TAC might also be explained by: (i) higher attentional demands, (ii) higher oculomotor demands, and (iii) the contour interaction effects present in the C-test (when using the crowded chart version).

As was mentioned in an earlier review, separate norm scores should be used when interpreting uncrowded and crowded acuity scores [6]. For children aged 6–10 years, crowding ratios of 2 and higher can be considered as increased [22]. In children aged 10 and older, crowding ratios of 1.5 can be seen as increased [40]. Studies report different findings with regard to the influence of viewing distance on crowding ratios. Dekker et al. reported lower near crowding ratios than distance crowding ratios, but did not have an explanation as for why this occurred. Huurneman et al. consistently found higher crowding ratios at near than at distance [27, 28]. These studies use different methods, but this difference does not offer an explanation for the observed finding. More research is needed to find out what mechanisms underlie differences between near and distance crowding ratios. In order to be able to detect vision problems due to crowding, it is of crucial importance to use sensitive screening charts. In general, sensitivity of visual screening tests can be improved by using flankers that are more tightly spaced and letter like with an optimal spacing of ~1.13× the optotype size [47].

During general school screenings, near vision is not a mandatory part of vision screenings, since there is evidence that poor distance vision has a higher prevalence than poor near vision [37, 43]. Therefore, near vision assessments do not seem to add value to general school screenings in children with normal vision and should be done when a child has reduced vision or experiences problems when reading or when doing near work.

Measuring visual acuity at a viewing distance chosen by the child can cause serious underestimations of near visual acuity in children (0.16 logMAR or 1.6 lines difference on a vision chart) [21]. An explanation for this is that under- or over accommodation of the lens can result in poorer acuities. In addition, it is difficult for the test leader to monitor viewing distance when the child is moving. There is substantial variability in viewing distances adopted across the studies that were included in this review. In older children, the majority of studies use a viewing distance of 40 cm. However, for the 3–6 year olds, viewing distance is as often 25 or 30 cm as it is 40 cm. With the exception of one study [21], there are no other studies systematically evaluating the effect of viewing distance with the same set of subjects, and therefore comparison between studies is not possible. Differences in near vision scores of 5 year olds for crowded optotypes [21, 26, 27] suggest that the 25 cm viewing distance could result in somewhat lower acuities, but since different children were tested across studies this comparison cannot be made. Viewing distance should be controlled for and, ideally, should be fixed and at 40 cm for near visual acuity because of the influence of accommodation, and in order to allow comparisons between studies. Experiments by Huurneman et al. show that children aged 4 years and older can very well respect the distance of 40 cm during near vision measurement [27].

Optotype choice has an impact on near visual acuities measured with letter scores resulting in lower acuity estimates than charts with symbols. This difference between acuity measured with letters and acuity measured with symbols was consistent across several studies and several age groups. Visual acuities measured with the 4AFC C-test and LEA symbols correspond well with each other [27, 28]. The ETDRS test presents 5 SLOAN letters per row with equal legibility, spacing between letters and rows is consistent; and size varies with (0.1) logarithmic intervals between lines. A limitation of the ETDRS chart is that letters cannot be read in countries where Roman characters are not used.

There were no consistent differences between distance and near visual acuities. Myers et al. did not find a difference between near and distance ETDRS acuity in children with normal vision [48]. Larsson et al. found no differences between near and distance VA in the first study, but did report better distance than near VA in the last study [33, 40]. The Fabian study reported no differences between distance and near [38]. In contrast with the Larsson studies [33, 40], Li et al. reported better near than distance VA [41]. Huurneman et al. found no clear near distance differences in children with VI [27, 28]. Boonstra et al. found better distance than near visual acuity in 3 ½–6 year old children with VI [21]. Finally, Dowdeswell et al., reported better near than distance visual acuity in children with NV while using the Bailey-Lovie chart at 0.3 and 6 m [23]. Collectively, the results of the included studies indicate that there no evidence for robust systematic differences between near and distance visual acuities in children with normal vision and children with low vision as long as the 40 cm distance at near is maintained during measurement.