Is vision screening in 3-year-old children feasible? Comparison between the Lea Symbol chart and the HVOT (LM) chart


Peter Jakobsson
Department of Neuroscience and Locomotion
Division of Ophthalmology
Linköping University
SE-581 85 Linköping
Tel: + 46 13 22 30 66
Fax: + 46 13 22 30 65


Purpose: The aim of this prospective study was to compare visual screening at the age of 3 years with screening at 4 years using two different charts.

Methods: A total of 478 3-year-old children were tested at four child health care centres (CHCCs). Of these children, 440 were tested again at the age of 4 years. A third group, a control group, consisting of 229 children, was examined only at the age of 4 years. All children were tested with both the HVOT chart and the Lea Symbol chart.

Results: Testability rates for 3-year-olds were almost the same with the Lea Symbol chart and the HVOT chart (82.8% and 84.8%, respectively). The corresponding rates for the same children tested at 4 years of age were 96.5% and 97.0%, and for the 4-year-olds not previously tested 92.9% and 92.8%. The mean testing time was somewhat shorter for the Lea Symbol chart in all three groups, but the difference was not statistically significant. The difference in the assessment of visual acuity between the two charts was small and less than 1/10th of a line. The positive predictive value was lower at 3 years (58%) than has previously been found at 4 years (74.6%).

Conclusion: Three-year-old children co-operate well in visual acuity testing. However, the examination time is a little longer and the testability rate is about 10% lower than at 4 years. Both 3-year-old and 4-year-old children can be tested equally well with the HVOT and the Lea Symbol charts.


In Sweden, screening of pre-school children has been in general use since the mid-1970s; in our earlier work we have shown that it is an effective method of detecting vision defects (Kvarnström et al. 1998). The attendance rate is more than 99%. The prevalence of deep amblyopia (visual acuity ≤ 0.3 or 20/60) is reduced from 2% to 0.2% with screening and treatment (Lennerstrand et al. 2000).

Despite this, 53% of children with amblyopia do not achieve a visual acuity (VA) of ≥ 0.7 (20/28) after screening and treatment (Kvarnström et al. 2001). Would earlier detection improve final VA? Reports have shown that treatment of amblyopia gives better results the earlier it is initiated (Nordlöw & Joachimsson 1966; Fulton & Mayer 1988; Flynn et al. 1999) and that treatment periods are shorter on average for younger children (Lithander & Sjöstrand 1991). However, later studies have shown that this is true only for severe amblyopia (Holmes et al. 2003), while moderate amblyopia is less influenced by age for children younger than 7 years (Pediatric Eye Disease Investigator Group 2002, 2003; Clarke et al. 2003). On the other hand, children with severe amblyopia (≤ 0.3) are a group of particular interest because not only do they benefit most from screening and treatment (Kvarnström et al. 1998), but they will become visually handicapped if they lose vision in the non-amblyopic eye (Jakobsson et al. 2002).

Visual acuity testing has been described as the best way to detect amblyopia (Köhler & Stigmar 1973). The problem lies in finding a reliable VA test for young children. There are no sensitive objective tests for detecting amblyopia and, therefore, subjective methods have to be used. For the last 20 years, the four-letter HVOT letter matching (LM) chart, developed by Anders Hedin, has been used in Sweden at child health care centres (CHCCs) at the age of 4 years (Hedin et al. 1980), and was consequently used in our earlier evaluation of the Swedish screening system (Kvarnström et al. 1998, 2001). The Lea Symbol chart was introduced in 1980 in Finland (Hyvärinen et al. 1980) but has not been used commonly in Sweden. It has not been compared with the Swedish HVOT chart and nor has it been tested on 3-year-old children with a follow-up at 4 years. Hered et al. (1997) compared the Lea Symbol chart with one variant of the HVOT chart and found the Lea Symbol chart to have a better testability rate at 3 years of age. Therefore, the purpose of this study was to investigate whether it is possible to reliably test 3-year-old children and to evaluate and compare the two charts in terms of testability and reliability.


All children born during a 1-year period and listed at five different CHCCs, three in Linköping and two in Huddinge, Stockholm, were included in this prospective study. The children were categorized into three groups. The first group, designated 3y, was tested at 3 years (± 3 months) and consisted of 478 children. Of these children, 440 were tested again at 4 years. The other 38 children were not re-tested as they had moved from the area. This group of 440 children was designated 3/4y and was tested with the same methods at 3 years and 4 years. A third group of 229 children, designated 4y, consisted of an additional group of unselected 4-year-old children attending their regular 4-year examination. They were not included in the 3-year screening, but were tested with the same methods as those in the other groups. This group functioned as a control group. The overall boys to girls ratio was 0.51 : 0.49. The children were tested by six nurses, who had many years' experience of VA testing. An orthoptist (GK) supervised the examinations and also took part in the testing.

Each of the three main groups − the children tested at 3 years of age (3y), the same children tested at 4 years of age (3/4y) and the children tested only at 4 years (4y) − were further divided into two groups, one of which was tested with the Lea Symbol chart first (Lea Symbol1), while the other was tested with the HVOT chart first (HVOT1).

Excluded from the study were children who needed or had parents who needed interpretation of the Swedish language and children who were already patients at an eye clinic.

The attendance rate was 98%, meaning that 98% of the children who were registered at CHCCs, in that age group, participated in the VA testing. Eleven parents/children refused to take part in the study, mostly because of lack of time. These children were excluded from the study. Some children entered the study at 3 years but failed the first test and, rather than submit to immediate retesting, the parents preferred to wait for retesting until the child was 4 years old. These children are included in the study as untestable because the testability rates were calculated on the first trial for all children.


The HVOT (Hedin et al. 1980) chart, which is widely used in Sweden for pre-school visual screening, has the four letters H V O T arranged in lines (Fig. 1). It differs from the Sheridan version of the HVOT test in that the distance between the letters is equal to the letter width, in order to better detect crowding (Rydberg et al. 1999). The Lea Symbol chart is well known internationally, and Finland, Norway and a number of states in the USA have accepted this chart for visual screening. The chart also has four symbols in lines. The symbols have been recalibrated to the international standard, Landolt C. The Lea Symbol chart and the HVOT chart are designed with a geometric progression of letter size. The Lea Symbol chart uses an exact geometric progression in steps of 0.10 logMAR unit, while the HVOT chart has a coarse geometric progression with increments varying between 0.09 and 0.18 logMAR unit. We used Lea Symbol chart no. 2504 as it fitted into the illuminator cabinets already in use in CHCCs in Sweden. Both charts use optotypes in lines because this is considered the best way to detect amblyopia with crowding. Single optotypes have not been used for screening in Sweden for the last 30 years and they are not used in clinical practice at the eye clinics.

Figure 1.

Anders Hedin's letter matching (LM) HVOT chart used for 4-year-old children in Sweden.

The time taken to perform this test was measured with a stopwatch and included the instruction time. The average testing time for one eye is depicted here and was calculated by adding the instruction time, the time for the right eye and the time for the left eye, and dividing this sum by two. All children were tested and instructed individually. All testing was performed at a distance of 3 metres from the chart. The eye not being tested was occluded with a facial patch and the right eye was tested first.

The children had a pointing card on a table in front of them in both methods, but could also give a verbal response. All children were tested with both methods; half of the group was tested with the HVOT chart first (HVOT1) and the other half with the Lea Symbol chart first (Lea Symbol1). The pass level was 0.65 (Lea 0.63), with less than two lines difference between the eyes. Four out of five optotypes had to be correct in the Lea Symbol test, and all except two optotypes in each line with the HVOT chart, according to the instructions for the respective method (with the exception of the first line containing three letters, where two had to be correct). As all lines except the first on the Lea Symbol chart contain five symbols, the pass criterion was 80% and was the same for all lines. The HVOT chart, on the other hand, contains a different number of optotypes on each line, ranging from three (0.1) to 12 (0.65–1.0). This resulted in the pass criteria for a line ranging from 50% to 83% with this chart. The nurses were instructed to test optotypes as small as the child could manage. The same procedure was performed at 4 years of age (both groups: 4y and 3/4y). The pass level at this age was 0.8 (20/25) and less than two lines of difference between the eyes. Children who failed either or both tests were re-tested a maximum of two times before being considered to have failed the screening. This was carried out with approximately 1 month between tests. If a child needed a re-examination, a copy of the two tests was given for training at home before the next visit. Children who failed both screening tests were referred to an eye clinic. Instructions were repeated at the re-examination.

A child who did not co-operate or who gave inappropriate responses was considered untestable, and a child who co-operated well but did not meet the VA criteria was considered testable, but to have failed the screening.


Testability rates at the first examination were almost the same for both charts, as can be seen in Fig. 2. There was no statistical difference between the methods at any age. However, there was a statistically significant difference between ages in the anova test. The testability was approximately 10% better at the age of 4 years and the t-test showed a significant difference (p < 0.002) between the 3y and 4y groups, and between the 3y and 3/4y groups, but not between the 3/4y and 4y groups. The most frequent reason was that the child refused to wear a patch. Children who had to be tested binocularly or with the parent's hand as an occluder were considered not testable.

Figure 2.

Testability rates by age group and chart used at the first visit. There were no significant differences between the two charts, but there was a significant difference between the 3y group and the other two groups.

The testability rate increased at the second examination of the 3-year-olds and 49% of the children who could not be tested the first time co-operated with at least one of the methods in the second session. This increased the testability rate at 3 years to 92%.

The mean testing times for one eye in the first examination, including instructions, were 2 : 21 min (minutes : seconds) (Lea Symbol) and 2 : 28 min (HVOT) in the 3y group; 1 : 48 min (Lea Symbol) and 1 : 50 min (HVOT) in the 4y group; and 1 : 21 min (Lea Symbol) and 1 : 29 min (HVOT) in the 3/4y group. In this comparison, Lea Symbol1 was compared to HVOT1 in order to avoid any training effects. The results are shown in Fig. 3. There was no statistical difference (anova) in the time taken for examination between the two charts. However, significant differences (p < 0.003) between the three age groups were observed for both methods. The largest difference was found between the 3y and 3/4y groups. At 4 years, children who had been tested at 3 years (3/4y group) had a significantly shorter test time (average 24 seconds) than those who had not been tested at 3 years (4y group).

Figure 3.

Mean testing time by age group and chart used. The figure shows the testing time for one eye, including instructions. The average total time for one child is twice this time. Times shown are for the first visit. There were no significant differences between the two charts, but there was a significant difference between all age groups.

Because each child was tested with both methods it was possible to compare the VA in each eye with the two methods. This comparison was performed for the first examination. Visual acuities of 0.32 and 0.63 in the Lea Symbol test were considered equal to 0.3 and 0.65, respectively, in the HVOT test. No-one had a VA of 0.25 in the Lea Symbol test and therefore no adjustment had to be made for this level of VA. The average difference between the Lea Symbol test and the HVOT test was less than 0.01 logMAR unit or less than 1/10th of a line (range − 0.43 to + 0.34 logMAR unit). There was a statistical difference between the Lea Symbol chart and the HVOT chart in the 3y and 4y age groups (p = 0.0398 and p = 0.0227, respectively), when the Lea Symbol chart was tested first. The HVOT test showed somewhat higher values, but the average difference was very small (0.010 logMAR unit and 0.007 logMAR unit, respectively). There was no statistical difference in the paired t-test between the first and second methods in any age group. The VA values at 3 years were compared with those at 4 years for the same children. The Lea Symbol test showed no statistical difference in this comparison, while the HVOT test gave on the average of one line (0.1 logMAR unit, p < 0.001) better acuity at 4 years.

Only one ‘missed’ child was found at the age of 4 years. This boy had a microtropia and should possibly have been detected at the age of 3 years. However, the boy was difficult to examine even at retesting and a third examination was planned, but the parents preferred to wait until the age of 4 years. He achieved full VA after treatment. No other missed cases were detected at the 4-year examination, but the actual number of false negatives cannot be calculated as 38 children had moved from the area and were not followed up.

Nineteen children in the 3y group did not pass the screening test. Eight of these were considered normal after examination at the eye clinic. The positive predictive value was thus 58% for the 3-year-olds. As the 4y group was not followed up, the positive predictive value could not be calculated for this group.


Visual acuity testing is an effective means of screening for amblyopia (Köhler & Stigmar 1973). Treatment of severe amblyopia has been shown to be most effective in younger children, which makes it important to detect this condition as early as possible (Lithander & Sjöstrand 1991; Holmes et al. 2003) and to find a method that is suitable for young children. There are no objective methods for measuring VA and the available subjective methods demand the child's attention and co-operation. Although there is an effective screening programme in Sweden, our previous study showed that 53% of children found to have amblyopia did not achieve VA ≥ 0.7 after screening and treatment (Kvarnström et al. 2001). The reasons for this are not known exactly, but one factor might be the age of detection. The aim of this study was to establish whether it is possible to reliably test 3-year-old children and to find out which chart would be the best at this age. We compared the Lea Symbol chart and the HVOT chart, which has been used on 4-year-old children in CHCCs throughout Sweden for about 20 years.

The testability rates in 3-year-old children were more than 80% with both the Lea Symbol and the HVOT charts, a figure that we consider acceptable for a screening test. The testability rate was better at 4 years, especially for those who had already been tested at 3 years. The reason is probably maturation with age rather than a training effect, as there was no statistical difference in this respect between the 4y group and the 3/4y group.

The most frequent cause for classifying a child as untestable was that he or she refused to wear a facial occlusion. The testability rate would have increased by almost 9% if the tests had not required an occluder patch. Many children accepted the parent's hand as an occluder but we classified these children as untestable because we consider the patching to be essential in a VA test. Salt et al. (1995) experienced the same problem with occlusion, especially in younger children. At 2.5–3 years, only 46% accepted occlusion, but at 3–4 years the percentage was 73% and at 4–5 years acceptance was 98%. To compare different types of occlusion was not an aim of this study and therefore no other types of occlusion, such as spectacles with an opaque shield over one eye, were tested. Hered et al. (1997) compared the Lea Symbol chart with another version of the HVOT test and generally found a better testability rate compared to our study, especially for the Lea Symbol test at 3 years. We were unable to verify this finding in our study. However, Becker et al. (2002) found a testability rate (76%) that was lower than ours using Lea Single Symbols for 3-year-olds and a comparable rate for 4-year-olds (95%).

The mean testing time for one eye in the first examination, including instructions, was 2 : 21 min (Lea Symbol) and 2 : 28 min (HVOT) at 3 years. This gives a total examination time for one child of 4 : 41 min and 4 : 56 min, respectively. It takes about 1 : 12 min more to test a 3-year-old compared to a 4-year-old. We view these figures as insignificant compared to those used for receiving and registering the child. The times found were longer than those presented by Hered et al. (1997), who recorded a test time of approximately 1 : 25 for 3-year-olds. However, their times did not include instruction time and they stopped testing when the child reached the pass level (0.5 or 20/40), while we continued to the child's maximum VA.

Our results from the second and third tests showed that both time and testability improved on each occasion.

There was no difference between the two charts with regard to VA in most of the age groups. This shows that they are equally reliable in this sense. This is in accordance with the findings of Rydberg et al. (1999), who found good agreement between the Lea Symbol test and the HVOT test. However, the VA values obtained with the Lea Symbol test at 3 years were more predictive of the values at 4 years, while the HVOT test showed higher values at 4 years.

The positive predictive value was 58% for the 3-year-olds. The positive predictive value for the 4-year-olds could not be calculated in this study, but a recalculation of figures from our earlier study (Kvarnström et al. 1998) gives a corresponding value of 74.6% for this age group. Although the number of children in this study is small and the data should be interpreted with caution, these data indicate that over-referrals are more common among 3-year-olds. Among the consequences of over-referral is the extra burden placed on eye clinics by unnecessary examinations.

Our conclusion is that 3-year-old children can be reliably examined with a VA test. A testability rate of approximately 80% permits mass screening, but the positive predictive value is lower than at 4 years. The testability rate and testing time did not show any advantages for either of the charts.


We wish to thank nurses Vivianne Eriksson and Gunnel Lassvik in Linköping, and Ann-Christin Johansson, Eva Jakobsson, Margareta Broman and Agneta Benzer in Huddinge, as well as head nurse Gunnel Wiklund.

The study was financially supported by the Sven Jerring Foundation, the 1st May Foundation, Frimurarna, Folkhälsoanslaget (LiÖ 50/98), the Margit Thyselius Foundation, KMA, the Lions Foundation, the Sigvard and Marianne Bernadotte Research Foundation for Children's Eye Care, Stiftelsen Solstickan, and Östergötlands Läns Landsting (95/092, 00/004).