Spectrum of ocular disease in children aged between 0 and 3 years at an Australian paediatric tertiary hospital

Background: Childhood ocular disease can be a significant health burden to the child, family and society. Previous studies have examined the spectrum of paediatric ocular disease presenting to tertiary hospitals; however, these studies have broader age ranges, smaller sample sizes, and are largely based in developing countries. This study aims to assess the spectrum of ocular disease in the first 3 years of life presenting to the eye department of an Australian tertiary paediatric hospital. Methods: The records of 3337 children who had their initial presentation at the eye clinic between the age of 0 and 36 months


| INTRODUCTION
Childhood ocular disease can be a significant health burden to the child, family and society. It is estimated that the global financial cost of childhood blindness is between US $6 billion and $27 billion, an unacceptably high figure. 1 Childhood blindness is a priority second only to adult cataract in terms of the number of blind person years lived and the consequent overall economic impact on the society. 2 About half the causes of blindness and visual impairment are potentially preventable or treatable. 3 Timely and periodic screening is critical for the detection of visual impairment and its aetiology to plan early intervention.
Although there are broad population-based reports on the eye health among Australian children, there exists no significant breakdown of paediatric ocular presentations, especially among infants. 4 There have been studies in other countries which have examined the spectrum of paediatric ocular disease presenting to tertiary hospitals, but none in Australia. [5][6][7][8][9][10][11][12][13][14] Furthermore, these studies have broader age ranges, smaller sample sizes, and are largely based in developing countries, which have a different spectrum of paediatric eye health compared to developed countries. In developing countries, vitamin A deficiency, corneal scarring from measles and infection are the major causes of visual impairment. 3 Other significant causes in all countries are congenital abnormalities, such as cataracts, glaucoma and hereditary retinal dystrophies. 3 The aim of this study is to assess the spectrum of ocular disease in the first 3 years of life presenting to the eye department of an Australian tertiary paediatric hospital. This study will supplement the level of detail missing in broader population-based Australian studies and will also provide a useful comparison of ocular presentations with other countries of different socioeconomic status. Furthermore, there will be a focus on the younger age group of 0-36 months, which is not present in comparable studies of other countries. Improvements in healthcare planning and management may then be implemented based on the results of this study to ensure the best care for these vulnerable infants.

| METHODS
A cross-sectional retrospective study was conducted at the Department of Ophthalmology in The Children's Hospital at Westmead in Sydney, Australia. The Children's Hospital at Westmead is a tertiary referral service for ophthalmology, providing services primarily to Western Sydney, in addition to the entire NSW state.
The records of 3337 children who had their initial presentation at the eye clinic between the age of 0 and 36 months were reviewed. Children were referred to the eye clinic through a variety of channels, including emergency department referrals, ward and specialty referrals, internal referrals such as electrophysiology testing and external community referrals. Records with incomplete or missing data were excluded from analysis, resulting in 3123 records included in this report. Data collection range spanned 6.5 years from 1st July 2012 to 31st December 2018. Ethics approval was obtained from the Sydney Children's Hospitals Network Human Research Ethics Committee. Strict confidentiality via a protected database was maintained.
Data collected included sex, age at initial presentation, diagnoses, visual acuity and method of visual acuity measurement. Classification of diagnoses was based on the international anatomical classification scheme established by Rahi and Gilbert,15 and included the site of primary abnormality, the primary diagnosis associated with this site, and a secondary diagnosis if more than one diagnosis was applicable. Additional sub-classifications were included to capture the granularity of presentations. Children who had diagnoses which did not fit in any anatomical site were grouped under 'other'. Furthermore, children who were found to have a normal ocular examination were described by their source of referral and categorised under 'referrals and screening'. Children were also classified as presenting with visual impairment or not, and the laterality of the visual impairment was also recorded. Visual impairment was defined as presenting with a visual acuity of less than 6/18 in the better eye in accordance with World Health Organisation international standards. 16 At times, this definition was difficult to apply due to the very young age of our sample group at initial presentation. However, in these cases, visual impairment was able to be identified from perusal of documentation and visual acuity from subsequent visits, paired with a diagnosis consistent with visual impairment. The inability to fix, follow and perform central, steady and maintained fixation at the initial visit was also taken into consideration.
Data was analysed using Jamovi (version 2.2.5, The Jamovi Project, Sydney, Australia). Evidence of normal distribution in age at presentation was assessed using the Shapiro-Wilk test. Age at presentation was grouped into 0-12, 13-24 and 25-36 months. Possible associations with age groups and sex were pursued using the Chi square test while strength of evidence was graded by the calculated probability value. A probability value of 0-0.001 was considered very strong evidence, a value of 0.001-0.01 was considered strong evidence, a value of 0.01-0.05 was considered evidence, a value of 0.05-0.1 was considered weak evidence and a value greater than 0.1 was considered no evidence. 17 Subset analyses were performed on the children presenting with visual impairment. Frequencies and percentages of variables were calculated, summarised and presented in tables.

| RESULTS
After records with incomplete or missing data were excluded, a total of 3123 records (93.6% of 3337) were included in this report (Table 1). There were slightly more boys (53.2%) than girls in this sample, however there were no evidence for differences between the sexes across age groups (χ 2 = 1.58, 2df, p = 0.453). A Shapiro-Wilk test was performed and there was very strong evidence that the distribution of presenting age departed from normality (W = 0.910, p < 0.001). Based on this outcome, and after visual inspection of the histogram of age revealed a positive skew (Figure 1), the median with the interquartile range were used to describe the variability of age. The median age was 8.0 months with an interquartile range of 12.0 months. The proportion of all children presenting with visual impairment was 10.3%, with 5.7% of all children presenting with bilateral visual impairment and 4.6% presenting with unilateral visual impairment.
There was no evidence that males were more likely to present with visual impairment than females (χ 2 = 0.716, 1df, p = 0.397). This was true within all age groups. Within the children that presented with visual impairment, there was no evidence that males were more likely to present with bilateral visual impairment than females (χ 2 = 0.0631, 1df, p = 0.802). This was also true within all age groups. Furthermore, within the children that presented with unilateral visual impairment, there was no evidence that males were more likely to present with right-sided visual impairment than females (χ 2 = 0.00739, 1df, p = 0.931). This was true within all age groups.
There was very strong evidence that children were more likely to present with visual impairment at an earlier age (χ 2 = 14.5, 2df, p < 0.001), as shown in Table 2. When split by sex, the strength of this evidence was more variable. There was weak evidence that males were more likely to present with visual impairment at an earlier age (χ 2 = 5.43, 2df, p = 0.066), while there was strong evidence that females were more likely to present with visual impairment at an earlier age (χ 2 = 10.6, 2df, p = 0.005).
Within the children that presented with visual impairment, there was strong evidence that younger children were more likely to present with bilateral visual impairment, while older children were more likely to present with unilateral visual impairment (χ 2 = 9.56, 2df, p = 0.008), as demonstrated in Table 3. Once again, when split by sex, the strength of this evidence was more variable. There was evidence that younger males were more likely to present with bilateral visual impairment than older males (χ 2 = 6.95, 2df, p = 0.031), while there was no evidence that younger females were more likely to present with bilateral visual impairment than older females (χ 2 = 4.40, 2df, p = 0.111).
Within the children that presented with unilateral visual impairment, there was no evidence that younger T A B L E 1 Characteristics of children who had their initial presentation at an Australian tertiary eye clinic between the age of 0 and 36 months. children were more likely to present with right-sided visual impairment than older children (χ 2 = 1.89, 2df, p = 0.389). This was true for both males and females. The frequencies of site of primary abnormality categorised by sex are presented in Table 4. Lids and lashes were the most common site of primary abnormality (11.3%), followed by conjunctiva and sclera (9.5%), and retina (7.0%). Nearly half of the children were referred for screening from other specialties and were found to have a normal ocular examination (48.9%). Children who had diagnoses that did not fit with the anatomical categories, such as idiopathic nystagmus, high refractive error and strabismic amblyopia, were grouped under "other" which accounted for 10.5% of presentations. High refractive error was defined as having more than five dioptres of refractive error as a myope or more than six dioptres of refractive error as a hyperope. 18 Delayed visual maturation was a diagnosis used to describe infants who did not F I G U R E 1 Frequencies of children who had their initial presentation at an Australian tertiary eye clinic between the age of 0 and 36 months.
T A B L E 2 Frequencies of children presenting with visual impairment categorised by age at initial presentation.

Age at initial presentation (months)
Presenting with visual impairment exhibit the ability to fix or follow objects in the environment, but subsequently improved by the age of 6 months without treatment. 19 There was no evidence for differences in sites of primary abnormality between the sexes (χ 2 = 10.7, 11df, p = 0.466). This was true within all age groups. The frequencies of site of primary abnormality grouped by age at initial presentation are presented in Table 5. Children who had a diagnosis which did not fit in any of the anatomical categories, such as idiopathic nystagmus, high refractive error and strabismic amblyopia. b Children who were found to have a normal ocular examination were described by their source of referral.
T A B L E 5 Frequencies of site of primary abnormality categorised by age at initial presentation.

Site of primary abnormality
Age at initial presentation (months) Children who had a diagnosis which did not fit in any of the anatomical categories, such as idiopathic nystagmus, high refractive error and strabismic amblyopia. b Children who were found to have a normal ocular examination were described by their source of referral.
Within the lids and lashes category, ptosis was the most common primary diagnosis followed by preseptal cellulitis. Nasolacrimal duct obstruction was the most common primary diagnosis within the conjunctiva and sclera category, followed by conjunctivitis. Within the retina category, retinopathy of prematurity was the most common primary diagnosis followed by retinoblastoma. Within the children that had a normal ocular examination, the most common cause for presentation was prematurity screening, followed by deafness referral and genetics referral. Strabismic amblyopia was the most common diagnosis within the "other" category, followed by high refractive error and delayed visual maturation. Subset analysis was performed on the children presenting with visual impairment. The frequencies of various diagnoses were categorised by laterality, sex and age at initial presentation (Table 7). There was no evidence for differences between the sexes with regards to the site of primary abnormality (χ 2 = 13.9, 9df, p = 0.162). This was true within all age groups. Lens abnormalities was the most common cause of visual impairment (21.4%), followed by retinal abnormalities (17.3%), and cerebral and visual pathways abnormalities (12.1%). Within the lens category, cataract was the only diagnosis. Retinoblastoma was the most common primary diagnosis within the retina category, followed by retinal and macular dystrophies. Within the cerebral and visual pathways category, structural abnormalities were the most common primary diagnosis, followed by hypoxic ischaemic encephalopathy. Once again, strabismic amblyopia was the most common cause of visual impairment in the "other" category, followed by high refractive error and delayed visual maturation. The six most common diagnoses overall in children presenting with visual impairment were cataract (21.4%), strabismic amblyopia (9.3%), retinoblastoma (6.5%), optic nerve hypoplasia (5.9%), high refractive error (5.9%), and cerebral and visual pathways structural abnormalities (4.0%).
Further subset analysis was performed on children presenting with unilateral visual impairment (Table 8). There was no evidence for differences between the sexes with regards to the site of primary abnormality (χ 2 = 9.13, 8df, p = 0.331). This was true within all age groups. Furthermore, there was no evidence for differences between laterality with regards to the site of primary abnormality. (χ 2 = 12.6, 8df, p = 0.126). This was true across sex and age groups. Lens abnormalities was again the most common cause of unilateral visual impairment (24.8%), followed by retinal abnormalities (17.9%), and whole globe and anterior segment abnormalities (15.2%). Within the lens category, cataract was the only diagnosis. Retinoblastoma was the most common primary cause of unilateral visual impairment within the retina category. Within the whole globe and anterior segment category, microphthalmos was the most common cause of unilateral visual impairment. Strabismic amblyopia was the most common cause of unilateral visual impairment in the "other" category, followed by high refractive error. The six most common diagnoses overall in children presenting with unilateral visual impairment were cataract (24.8%), strabismic amblyopia (18.6%), retinoblastoma (9.7%), microphthalmos (5.5%), other optic nerve abnormalities (4.1%) and anterior segment dysgenesis (3.4%). Children who had a diagnosis which did not fit in any of the anatomical categories. b Children who were found to have a normal ocular examination were described by their source of referral.
T A B L E 7 Sites of primary abnormality in children presenting with visual impairment categorised by laterality, sex, and age at initial presentation.

| DISCUSSION
In this study, the spectrum of ocular disease in children between the age of 0 and 36 months presenting to an Australian tertiary paediatric hospital was evaluated.
There have been analogous studies examining the spectrum of ocular disease at a paediatric tertiary hospital in Africa, the Middle East and India, albeit with broader age ranges and smaller sample sizes. [5][6][7][8][9][10][11][12][13][14] To our knowledge, this is the first study of its kind in Australia and more widely in the developed world. Strikingly, there is an absence of studies structurally analogous to ours from the European, Pacific or American regions. Although there are numerous studies which examine the causes of childhood visual impairment, 15,[20][21][22][23] there is a deficiency of studies which examine the full spectrum of presenting ocular conditions in a paediatric tertiary hospital. This may be due to the presence of established population-based systems in place to monitor childhood ocular morbidity. For example, in Australia, the Medicare Benefits Schedule provides information on health services for which benefits may be paid, including optometric and ophthalmological services. 4 General practice activity is captured by the Bettering the Evaluation And Care of Health (BEACH) program, which is a continuous national study taken from a random sample of general practitioners across the country. 4 The National Hospital Morbidity Database and Australian Congenital Anomalies Monitoring System provide further information on childhood ocular disorders in hospitals. 4 In the state of New South Wales (NSW), the State-wide Eyesight Preschooler Screening (StEPS) program has offered free vision screening to all 4-year-old children since 2008. 24 The screening program is conducted in preschools, childcare centres and other children's services, with the aim of detecting vision problems prior to children starting school, at an age when reliable vision testing can be achieved and treatment for childhood ocular conditions is more effective than when they are older. The StEPS program model is unique in Australia and internationally and is one of the largest, most systematically implemented and evidence-based vision screening programs available. It is a highly appropriate and effective strategy for guiding young children to early intervention and treatment for childhood ocular conditions, with an estimated 96.4% of 4-year-old in NSW having been offered the opportunity to be screened through StEPS. 24 Our study is the first study to provide detailed data on vision impairment in the under 3-year-old cohort. Current vision screening guidelines include an assessment at 6 weeks with the new born program and then at 4 years. Our data show that 1647 (49%) of cases present before 8 months of age. This suggests that vision should be assessed around 6-8 months of age. Our work provides foundation evidence for more detailed investigation into adding an additional vision screening timepoint.
The socioeconomic impact of low vision and blindness from paediatric eye disease in Australia is substantial. A recent analysis by Grigg et al. in conjunction with Deloitte Access Economics revealed that there are almost a third of a million (332 936) Australian children with visual impairment or the potential to become visually impaired in Australia in 2015. 25 Total health costs to treat children with diseases of the eye and adnexa are an estimated $439 million in 2015, or 11.3% of the total health system expenditure on eye conditions in 2015. The total cost of disability adjusted life years is estimated to amount to $1.31 billion, or $3880 per child with visual impairment in 2015. 25 T A B L E 7 (Continued) For all the benefits of these Australian populationbased systems, only the broadest snapshot of ocular disorders is captured. For example, ocular motility, adnexal disorders and head injuries are the leading diagnoses for hospitalisations, yet there is no information on outpatient hospital clinic presentations. 4 Our study aims to fill this T A B L E 8 Sites of primary abnormality in children presenting with unilateral visual impairment categorised by laterality, sex, and age at initial presentation.  gap in depth and granularity of data, thus providing a level of understanding which will be useful for proper health care planning and management. There was very strong evidence that children were more likely to present with visual impairment at an earlier age. Furthermore, within the children that presented with visual impairment, there was strong evidence that younger children were more likely to present with bilateral visual impairment, while older children were more likely to present with unilateral visual impairment. To our knowledge, this finding has not been reported elsewhere for the age range used in our study. The likeliest reason for this finding would be that caregivers would be most likely to notice abnormal visual behaviour at an earlier age in a child with bilateral visual impairment than a child with unilateral visual impairment, as the child would not be compensating with a well-functioning eye.
The prevalence of all children presenting with visual impairment in our study was 10.3%. This finding is similar to the 8.8% figure found in the study by Demissie et al., who studied 358 children aged up to 16 years old over a period of a year at an Ethiopian tertiary teaching hospital. 6 However, important contextual differences between the studies include the socio-economic status of the society, age range and sample size.
There was no evidence for differences in the prevalence of any diagnoses between the sexes across all age groups in our study. This finding is echoed in the study by Pineles et al., in which the insurance data of nearly 11 million children up to the age of 19 were examined in order to define the prevalence of medical eye disease diagnoses among children enrolled in commercial insurance plans in the United States. 26 There was also no evidence found in our study for differences in the prevalence of visual impairment between the sexes.
For all children in our study, the three most common sites of primary abnormality were lids and lashes (11.3%), followed by conjunctiva and sclera (9.5%), and retina (7.0%). In the children who presented with visual impairment, the three most common sites of primary abnormality were lens (21.4%), followed by retina (17.3%), and cerebral and visual pathways (12.1%). Studies of childhood visual impairment in the developed world have yielded similar results. 15,22,23 In Turkey, Ozturk et al. found that cerebral visual impairment was the most common sites of childhood visual impairment, followed by retina and lens abnormalities. 23 Boonstra et al. in the Netherlands also found cerebral visual impairment, retina and lens abnormalities as the most common categories of visual impairment. 22 Although the three most common causes of childhood visual impairment remain consistent between these studies and ours, the hierarchy is different. This may be explained by differences in age range, as our study focussed on children aged less than 36 months, whereas other studies included children up to 21 years of age. Additionally, as our study captured data from only one centre, there may have been children that were not referred to our centre or managed elsewhere.
Interestingly, in our study, only 7 out of 156 children who had retinopathy of prematurity as their primary diagnosis developed visual impairment, likely as a result of a robust screening and treatment programs. This is supported by the fact that prematurity screening is our most common presentation in the referrals and screening category.
Our study has some limitations. First, it is a retrospective study. Temporal relationships may be difficult to assess, and our study can only determine association and not causation. Second, although our study was conducted at the largest paediatric tertiary teaching hospital, it is nevertheless limited to one centre. Multi-centre studies would improve the quality and quantity of data collected. Third, our study is limited to the hospital population and is not reflective of the general population. Fourth, it represents the child's initial presentation and does not reflect the full workload of the department as determined by ongoing management.
The study strengths include the following. First, it scrutinises a very young age range with a large sample size which is rarely seen in other studies. Second, it is conducted in a high-income developed society, which differs from studies with a similar structure, and therefore, fills in knowledge gaps. Third, it contains a high level of detail with regards to the breakdown of presenting conditions. Fourth, although our study does not reflect the full workload of the department, it is illuminating on workforce requirements and areas of training for the most common presentations.

| CONCLUSION
The spectrum of ocular disease in children who had their initial presentation at an Australian tertiary paediatric hospital between the age of 0 and 36 months was assessed. The majority of children had their initial presentation at an earlier age. Children were more likely to present with visual impairment at an earlier age. Bilateral visual impairment was more common in younger children, while unilateral visual impairment was more common in older children. Lids and lashes were the most common site of primary abnormality, followed by conjunctiva and sclera, and retina. The most common primary diagnoses overall were strabismic amblyopia, retinopathy of prematurity, and nasolacrimal duct obstruction. In children with visual impairment, the most common sites of primary abnormality were lens, retina, and cerebral and visual pathways. The most common primary diagnoses in children with visual impairment were cataract, strabismic amblyopia and retinoblastoma. Workforce requirements and areas of training should be tailored to the above findings.