Childhood cerebral visual impairment often goes unrecognised, especially in case of mild brain damage which has remained obscure, delaying rehabilitation to support normal development. Dutch youth health care physicians (physicians providing preventive services), paediatricians and paediatric ophthalmologists indicated difficulties in recognising higher visual problems and expressed the need for clear diagnostic guidelines, whereas experts of low vision services indicated that it is not efficient to perform time-consuming assessments of cerebral visual functions, e.g. object and face recognition, in all referred children. They expressed the need for a quantitative, evidence-based measure that could help with the selection.1

Hård et al.2 suggested that ‘crowding’ might be an indicator for visual perception problems. Crowding is the phenomenon of impaired discrimination of visual stimuli in the middle of other stimuli.3 Abnormal crowding may show in daily life as difficulties handling complex visual situations, for example finding a toy in a toy box or coping with busy environments like supermarkets; reading may be hampered, as well as visual acuity testing with routinely used cards.4 It may either concern a selective attention problem, negatively influencing specific higher perceptual functions,5 or it might be a failure in the neural development of small integration fields.6 In both instances, crowding might predict (unidentified) brain damage and, indirectly, visual perception problems.7–9

Children without any neurological impairment who have developmental visual problems, like strabismic amblyopia, are also at risk of crowding10,11 as well as of cerebral visual impairments, for example motion-defined form perception.12 All these visual problems in both children with amblyopia13 and children with cerebral visual impairment are associated with anatomical and functional abnormalities of the brain's visual system.13,14

If, indeed, abnormal crowding is more common in children with brain damage than in other children, a small addition to routine visual acuity assessment might improve the detection of children at risk of cerebral visual impairment. Instead of a single acuity test, two tests at the same test distance would have to be performed, namely a test with a single optotype and a test with multiple optotypes, after which the difference can be expressed as a ratio. Studies comparing grating acuity (Teller Acuity Cards) with optotype acuity (single and/or multiple optotypes) also indicate that larger differences between acuities are found in children with strabismic amblyopia11 and children with brain damage15 than in normally developing children and children with ocular disorders.11,15

Therefore, the first aim of this exploratory study was to assess whether the crowding ratio, estimated by the Cambridge Crowding Cards (optotypes), and the ratio between grating acuity and optotype acuity, differentiate better than visual acuity alone between children without any neurological and ocular abnormalities, children with ocular disorders only, and children with (indications of) brain damage.

In clinical practice, different levels in the care supply chain, i.e. youth health care physicians or nurses, ophthalmologists, and experts in low vision services, see different populations and have different detection goals and indications for referral.

Therefore, our second aim was to study which parameter (visual acuity alone, different ratios) for each level has the best sensitivity and specificity to meet their detection goal.

For this cross-sectional study, a heterogeneic group of 107 children in the age range 3–12 years was recruited in the Rotterdam area through primary schools, two rehabilitation centres, and the Rotterdam Eye Hospital. For all participants, informed consent was obtained from their parents or guardians.

Medical files and the children's histories were used to categorise participants as ‘typically developing without indications of neurological or ocular abnormalities’ (group 1), ‘ocular abnormalities without indications of neurological abnormalities’ (group 2) and ‘(indications of) brain damage with or without ocular abnormalities’ (group 3). Cerebral palsy (CP) as well as visual problems not explained by ocular abnormalities were considered indications of brain damage. Children without neurological symptoms with high myopia or hyperopia (spherical error≥6D) were included in group 2, and those with smaller refractive errors in group 1.

To meet study aim 2, we had to compose different control and target groups per discipline. In order to refer properly, youth health care physicians want to identify all children with visual problems irrespective of their cause (target: groups 2 and 3, controls: group 1), whereas ophthalmologists and specialists at low vision services want to differentiate between children at risk of cerebral visual impairment (target: group 3) and children with ocular abnormalities only (controls ophthalmologist: group 1 and 2, controls low vision centre: group 2). To this end, the following child characteristics were collected from the clinical files: visual acuity, refractive errors, use of glasses or lenses, strabismus, contrast sensitivity, other ocular abnormalities, and if present, aetiology of brain damage. Visual fields were checked using the confrontation test. Although refractive errors and strabismus are associated with crowding problems, these children were not excluded from the analyses, because this reflects clinical practice.

The study adhered to the Declaration of Helsinki and the protocol has been approved by the Medical Ethical Testing Committee of the Erasmus University Medical Center Rotterdam (MEC nr: 2006-056).

In the period June 2006–September 2010, visual acuity was assessed in all participants with the Cambridge Crowding Cards16 to assess uncrowded and crowded optotype acuity, and the Teller Acuity Cards-II17 to assess grating acuity. The uncrowded Cambridge Crowding Cards show a single letter and the crowded cards show a target letter surrounded by four letters, with a distance between letters of half a letter width. To minimise the effect of differences in test distance, the Teller Acuity Cards as well as the Cambridge Crowding Cards were presented at a distance of 3m. Outcomes of the Teller Acuity Cards were corrected for the greater than recommended presentation distance, using the following formula: cycl/deg=(test distance/57)×cycl/cm. Children were assessed binocularly, if applicable wearing their own glasses. Grating acuity was assessed first, followed by single and crowded optotype acuities. During Teller Acuity Card assessment, children were asked to actively point out the location of the stripes (left or right), and during optotype assessment, they had to name or match the target letter. All tests were started with coarse gratings/large letter sizes. Finer gratings/smaller letter sizes were presented until the child was unable to identify the correct position/letter. Maximally three cards were presented per visual acuity level and the visual acuity threshold was defined as the highest visual acuity level with two out of three correctly identified grating positions/letters. All assessments were performed by trained orthoptists.

To enable comparisons between grating and optotype acuities, we expressed all acuities in cycl/deg. We determined two cut-off values for crowded visual acuity, because in youth health care, children with ‘subnormal vision’ should be referred for ophthalmological examination, whereas ‘low vision’ is a traditional eligibility requirement for low vision services. Subnormal visual acuity was defined as a crowded acuity lower than 20cycl/deg (Snellen equivalent 6/9) for 3½–4½ year olds and lower than 30cycl/deg (Snellen equivalent 6/6) for older children (norms from The Cambridge Crowding Cards leaflet), whereas low vision was defined as a crowded acuity equal to or lower than 10cycl/deg (Snellen equivalent 6/18).18

Crowding was quantified as a crowding ratio (CR), which is defined as the acuity presenting a single optotype, divided by the acuity presenting several optotypes (linear optotype acuity). Because linear acuity tests mostly result in lower visual acuities than single optotype testing, the crowding ratio is usually higher than 1.16 In typically developing children, there is a developmental effect, with a mean crowding ratio (CR 1.8) in 3–4 year olds, decreasing with age.19 Therefore, a crowding ratio≥2 (i.e. halving of the visual acuity outcome) was considered as the standard for abnormal crowding, irrespective of age or developmental level.

The following ratios were calculated for each participant: crowding ratio (uncrowded/crowded optotype acuity), grating acuity/uncrowded acuity, grating acuity/crowded acuity. Mean (SD) and median ratios were calculated for the three participant groups. Between-test differences were analysed with Friedman and Wilcoxon Signed Rank tests and between-group differences with Kruskall–Wallis tests.

To answer the second study question, per specialist discipline, the proportion of correctly identified children-at-risk within the target group (sensitivity) and the proportion of correctly identified non-targets within the control group (specificity) were calculated. We plotted ROC-curves for crowded visual acuity and each ratio, to check which outcome and cut-off value could be used best by each discipline. If visual acuity, measured during the study, appeared different from the received information, the child was not excluded from the analysis or transferred to a different group, because we wanted to act on clinically defined risks. To be certain that the presence of abnormal crowding could not be explained by the presence of strabismus, risk factor for amblyopia, we calculated the correlation (Spearman's ρ) between a crowding ratio ≥2 and manifest or latent strabismus in the total study population.

This study explored an opportunity to improve clinical identification of children at risk of cerebral visual impairment, based on visual acuity ratios, in order to advance effective referrals for specialist neuropsychological assessment. The outcomes indicate, that the crowding ratio (ratio between uncrowded and crowded acuity) has a favourable sensitivity (67%) and specificity (79–86%), independent of visual acuity, to distinguish children at risk of cerebral visual impairment from children with ocular pathology only. Current referral for specialist low vision assessment, based on visual acuity, is less sensitive and specific for this aim: the ‘low vision’ criterion only detects 29% of the at risk children. A crowding ratio ≥2, easily checked by orthoptists, might be a valuable referral criterion. However, for screening of unselected groups of children, such as in youth health care, it does not provide useful information. Youth health care professionals best continue applying subnormal crowded acuity as an indication for referral.

The present study indicates that the Cambridge Crowding Cards are highly applicable (80–100%) and better accepted than the addition of a grating acuity test to routine assessment. Further studies have to prove whether or not ratios differ using different crowded acuity tests, because acuity tests vary in letter distances and step sizes (Snellen vs. log-based acuity). A published study in 3-year olds comparing results of different crowded acuity tests, indicates that differences can be expected,21 although these differences might have been age-dependent.

Even though crowding has been extensively studied during more than 80 years22 and is considered clinically relevant, little is known about developmental changes and normal limits in children. We were surprised to find that 14 out of 59 typically developing children had crowding ratios≥2. This might be explained by the relatively large number of younger children in this group: a significant relationship between crowding ratio and age in the group without brain damage suggests a developmental phenomenon.

Because in an earlier study applying Cambridge Crowding Cards in children without ocular and neurological abnormalities16 standard deviations have not been reported, we cannot statistically compare findings. Mean crowding ratios found in 3–4 year olds seem comparable (1.7 vs. 1.8), whereas the mean crowding ratio in 5–7-year olds in the present study is slightly higher than in the earlier report (1.6 vs. 1.2).

Although the use of crowding ratio≥2 may help to detect children with unconfirmed brain damage, especially those with subnormal vision that are otherwise not referred for or entitled to specialised diagnostics and low vision care, we did not address the question whether or not the crowding ratio directly predicts perception problems and can be used as selection criterion in low vision services. Studying the developmental changes of the crowding ratio in relation to results on tasks with a high attentional load or with stimuli that can be considered ‘crowded’ (i.e. visual search tasks and object perception tasks under suboptimal conditions), might help to answer the question whether or not the crowding ratio can be used as selection criterion and predict higher visual impairments. If target identification in ‘crowded’ stimuli is modulated by attention and determined by the attentional resolution,5 task performance might be dependent on age and task duration. Young children with brain damage might have a higher crowding ratio as well as weak perception performances, whereas older children might perform normally on the short acuity task but worse on perception tests, with their longer duration i.e. with a higher attentional load.

This study was financed by Royal Dutch Visio and the Intellectual Disability Medicine research group of the Erasmus University Medical Center Rotterdam.

The authors have no conflicts of interest to declare.