The online questionnaire was created using the 51-question inventory by Dutton (2010), that explores difficulties experienced by children in everyday tasks that are vision-dependent. The questionnaire used a scale of qualitative responses; “Never” (score (1), “Rarely” (2), “Sometimes” (3), “Often” (4), and “Always” (5). An option of “Not Applicable” was available and if selected, was removed from the analysis. A positive result (a score of 4 or 5) on three or more of questions 2, 18, 19, 24, and 27 was used as a positive screening for suspected CVI. This 5-question screening tool was determined by those difficulties commonly reported in children with CVI and rarely in those without (Dutton, 2010) and is a reliable diagnostic screening tool (Macintyre-Beon et al., 2012; Philip et al., 2016) with a good construct validity; sensitivity of 81.7% and specificity of 87.2% (Gorrie et al., 2019).

Retrospective review of participants' clinical records was conducted and eight factors which could impact on the incidence of behavioural features of CVI were identified: age, gender, corrected visual acuity (binocular LogMAR), ametropia (best vision sphere of the least ametropic or fixing eye), magnitude of astigmatism (of least astigmatic or fixing eye), strabismus (present or not), accommodation (accurate or not), nystagmus (present or not).

This study was part of a wider longitudinal study in children with DS and had ethical approval from the National Institute for Social Care and Health Research Ethics Service (08/MRE09/46, amendment 5, 7th July 2016).

A response rate of 36.7% was achieved; 81 responders. The gender and age of the respondents were compared to those of all invited participants. Gender (55.6% female) and age (4.4–17 years, mean 9.9 years) did not differ significantly from the invited population (χ² = 0.86, p > 0.05 and t = −1.53, p > 0.05, respectively).

Participants had a mean binocular visual acuity of 0.29 LogMAR (standard deviation, SD = 0.19), recorded using a variety of tests, based on the child's age and abilities: Cardiff Acuity Test, Crowded or Uncrowded LogMAR Kay Picture Test, Crowded or Uncrowded LogMAR Kay Acuity Test and the Keeler Crowded or Uncrowded LogMAR Visual Acuity Test.

Ten children (12.3%) in the study population had nystagmus and 13 children (16%) manifest strabismus (10 esotropia including 6 fully accommodative, two exotropia, and one vertical tropia).

Accommodative ability was recorded using dynamic retinoscopy; 39 children accommodated accurately, 39 had under-accommodation (all wore multifocal spectacle correction) and three children had inconclusive results.

Choice of refraction method was based on clinical need: static, Mohindra or cycloplegic retinoscopy. The best vision sphere of the least ametropic eye (fixing eye in strabismus) was recorded. There was no correlation between ametropia and age; Pearson, r = 0.04; n = 81; p > 0.05 (p = 0.72). The distribution of refraction is shown in Table 1.

Of the 81 children, 31 screened positive for suspected CVI; a prevalence of 38.3%. This report uses the term suspected CVI to mean a positive classification according to the five-question criteria.

The raw total score for each participant was attained by summing all question responses and was expressed as a percentage of the total questions applicable for that participant. Both groups were normally distributed (Figure 1). The minimum score would be 20% (every question recorded as “Never” and awarded one). No child had a minimum score. The mean total score for children with suspected CVI was 59.5% (SD = 10.5%, range = 41.2–78.8, Shapiro-Wilk p > 0.05) and the mean for children without suspected CVI was 44.1% (SD = 10.3%, range = 24.0–67.8, Shapiro-Wilk p > 0.05); the difference was significant (t = 6.286, p < 0.001).

The positive score for each participant was attained by summing the number of positive responses (Often or Always) and expressed as a percentage of the total applicable questions. Of children with suspected CVI, the mean number of positive responses was 36.7% (SD = 17.7%, range = 8.3–76.5%), compared to 13.5% (SD = 11.1%, range = 0–47.1%) among children without suspected CVI. The difference was significant (t = 7.269, p < 0.001). The range of positive responses is shown as a continuum in Figure 2, illustrating the large overlap between those children with and without suspected CVI.

Figure 3 shows the number of participants whose parents responded positively to each question and therefore highlights the questions that most frequently elicited positive responses and the weighting of suspected CVI and non-suspected-CVI responses for each question.

The 10 questions most frequently eliciting positive responses (“Often” Or “Always”) were “Does your child”:

2. Have difficulty walking downstairs?

15. Have difficulty seeing things which are moving quickly, such as small animals?

18. Have difficulty catching a ball?

19. Have difficulty seeing something which is pointed out in the distance?

26. Sit closer to the television than about 30 cm?

27. Find copying words or drawings time-consuming and difficult?

29. Find uneven ground difficult to walk over?

37. Find it difficult to keep on task for more than 5 minutes?

38. After being distracted, find it difficult to get back to what he or she was doing?

43. Have difficult behaviour in a busy supermarket or shopping centre?

There were several questions that elicited positive responses equally between children who met the criteria for suspected CVI and those who did not (e.g., 47; Does your child mistakenly identify strangers as people known to them?) and some questions that were responded to positively by large numbers with and without suspected CVI (e.g., 27; Does your child find copying words or drawing time-consuming and difficult)?

There were five questions that elicited responses only from children with suspected CVI, although numbers were small:

8: Does your child leave food on the right or left side of their plate?

12: Does your child bump into door/frames or partly open doors?

45: Does your child have difficulty recognising close relatives in real life?

46: Does your child have difficulty recognising close relatives from photographs?

51: Does your child have difficulty recognising familiar objects, such as the family car?

Each factor was tested independently to determine any association with positive screening outcome for CVI. Continuous factors were tested using a Mann Whitney U-Test; categorical factors with a χ² test. The raw data for each categorical factor were inspected for outliers. Two data points with refractive errors of −12.00 D and −7.25 D were more than 1.5 × IQR away from the mean and were removed from the analysis.

Of these eight factors, only absolute refractive error (equivalent sphere) was found to be significantly different between those with and without suspected CVI (p = 0.010) with two outliers removed. The higher the refractive error, the more likely was a child to screen positive for CVI. To determine whether refractive error is a suitable predictor for the incidence of suspected CVI, a univariate logistic regression was performed. The outcome of the logistic regression is given in Table 2. However, the model was only able to correctly predict suspected CVI in 62% of cases, compared to the null model's 60.8%.

Male predominance has previously been recognised amongst the DS population and the male to female sex ratio in children with DS has been reported as 1.23 (Kovaleva et al., 2001). Although the current analysis sample has an uncharacteristically high proportion of females, there does not appear to be any gender bias in responders and gender does not appear to be influential on suspected CVI outcome.

The relatively high prevalence of nystagmus, reduced acuity, strabismus and accommodative deficits reported here are consistent with other studies. Analysis showed that none of these functions was predictive of suspected CVI and they are therefore unlikely to be causal factors of the behavioural features. A recent study of children with a diagnosis of CVI and with a variety of risk factors (not including DS) reported that almost half had normal visual acuity (Sakki et al., 2021), confirming that CVI should never be assumed to be limited to children with poor acuity (Dutton, 2021).

Longitudinal studies have demonstrated a failure of the emmetropisation process in children with DS and that large refractive errors tend to either remain stable or increase with age (Haugen et al., 2001; Cregg et al., 2003) in young children, although no large changes to the spherical component of refractive error occur over the age of 4 years (Al-Bagdady et al., 2011). Thus, refractive errors in this study's population of 4–17-year-olds can be considered stable and this was confirmed by Pearson correlation.

Absolute refractive error was the only optometric factor to be significantly different between those with and without suspected CVI. The odds of screening positive for CVI increase by approximately 40% per dioptre of absolute spherical equivalent and children with over 5D of refractive error are more likely to screen positive for CVI than not. However, a model based on refractive error increases the likelihood of correctly predicting suspected CVI by only a small amount. The association between CVI and refractive error among other groups of children does not appear to have been investigated.

Although effort was made to select a random sample, some level of bias is likely present since all invited children were existing patients of Cardiff University's Special Assessment Clinic and may have already been experiencing some sort of visual difficulty. Information such as ethnicity, level of education and income were not collected. No participant had a diagnosis of brain injury, but childhood medical history was not obtained; this could be informative in terms of exposing the underlying cause of behavioural features, such as subtle oxygen deprivation.

It is clear that visual perceptual difficulties are common amongst children with Down syndrome and that further work needs to be done to understand the origin of the difficulties. Visual perception difficulties have been identified as a potential cause for academic underachievement (Williams et al., 2011), and as children with DS are considered visual learners (Yang et al., 2014) recognition of possible CVI is important in ensuring that these children can access education tailored to their requirements (Dutton, 2021). The findings in the current study would suggest that generally, children with DS tend to exhibit more problems with visual perception than might be expected.

Cognitive impairment is a characteristic of Down syndrome and many of the problems occurring in children both with and without suspected CVI may be attributable to the cognitive impairment. However, the impact of CVI on performance on tests of cognitive ability, which often involve the use of visual attention, spatial orienting, visual perception and visual motor skills (Moore et al., 2002) has not, to our knowledge, been explored. If CVI is present and unrecognised, it may be that a child's cognitive function is at risk of being underestimated.

The majority of children with DS aged 4–17 years in this study experienced some level of visual perceptual difficulty and 38.3% met the screening criteria for suspected CVI.

Whilst children with DS are known to have a high prevalence of visual deficits, this study has shown that only refractive error is an indicator of the likelihood of CVI-related behavioural features. It appears likely that CVI is the explanation for the frequent visual perception impairments in children with DS and that further investigations are warranted. Optometric deficits are unrelated conditions that often coincide within this group.

Further research is clearly warranted into the aetiology of the visual perceptual problems that appear so prevalent in children with DS, and the likelihood of a diagnosis of CVI. According to Sakki et al. (2021) “the economic, social, and personal burden of CVI is high, with adverse effects of coexisting disorders increasing the burden further.” _It is essential that visual problems associated with CVI are explored in the assessment of children with DS, and that difficulties are not simply considered to be due to the learning disability or to inappropriate behaviour. Targeted strategies can be helpful in ameliorating the effects of CVI (McKillop and Dutton, 2008; Tsirka et al., 2020), and these should be considered for any child with DS who exhibits difficulties.