Blindness is decreasing among children born preterm during the last four decades in Denmark

Children born preterm are believed to be at increased risk of visual impairment (VI). The increased survival rate of extremely preterm children may have changed the spectrum of diseases occurring postnatally. The aim of this study was to analyse the prevalence and causes of VI in an ex‐preterm Danish population during the last 4 decades.


| I N T RODUC T ION
Preterm (PT) children have a high risk of decreased visual function due to ocular and systemic complications to prematurity (Deng et al., 2019;Leung et al., 2018).Each year there are 61 000 livebirths in Denmark, 6% of those are preterm (PT) defined as a pregnancy lasting for less than full 37 weeks of gestational age (GA) and 1% are born <32 GA (Paris-Horan et al., 2022;Sundhedsdatastyrelsen, 2022aSundhedsdatastyrelsen, , 2022b)).Preterm birth is subcategorized in extreme (GA < 28), very , and moderate prematurity   (WHO, 2023).The standard neonatal intensive care has improved over the last decades, resulting in an increased survival rate of extreme and very PT (Cueto et al., 2022;Hasani et al., 2023).A shift in the occurrence and the spectrum of diseases appearing postnatally may therefore be suspected.
Systemic complications to prematurity can be caused by dysregulation of growth hormones and oxygen supply, due to an early exposure to external environments, compared to the natural regulations inside the uterus (Scratch et al., 2015;Serenius et al., 2013).Asphyxia during labour and postnatally due to immature lungs, and an increased risk of cerebral haemorrhage can lead to brain damage in PT children (Chokron & Dutton, 2023).In such cases, the anatomy of the eye can be normal but visual acuity is low due to damage of the retrochiasmal visual pathways or higher visual functions, that is, cerebral visual impairment (CVI) (Dutton, 2010(Dutton, , 2013)).
Ocular complications to prematurity include retinopathy of prematurity (ROP) and sequelae to treatment for ROP, optic nerve injury due to ischemia or cerebral lesions and arrested foveal development (Chiang et al., 2021;McLoone et al., 2006;Sjöstrand & Popović, 2020;Venkataraman et al., 2022;Wenner et al., 2023).ROP is a retinal disease, only seen in PT children and is caused by a disturbance in the oxygen supply to the immature peripheral retina.In mild stages the disease can regress spontaneously, but severe stages may lead to retinal detachment and blindness if not treated.Laser therapy for ROP was introduced in the early 1990s and can affect the visual function by reducing the visual field (Holmström & Larsson, 2008;Hunter & Repka, 1993;Larsson et al., 2023).Arrested foveal development is characterized by lack of foveal pit depression and continuation of inner retinal layers crossing the fovea.The lack of a foveal pit is more pronounced with decreasing GA and does not evolve with increasing post menstrual age (Sjöstrand et al., 2017).The consequences of arrested foveal development can be decreased visual acuity, and it is seen in PT with and without ROP (Bowl et al., 2019;Sjöstrand & Popović, 2020;Venkataraman et al., 2022).
In Denmark all children with blindness and visual impairment (VI) under the age of 18 years are registered in The National Register of Blind and Visually Impaired Children (NRVIC), of whom 8.5% of the registered children are born <32 GA (Paris-Horan et al., 2022).The register was established around 1980 and is regularly updated with clinical information (Kann, n.d.).Health care professionals are required by law, given parental consent, to inform the NRVIC of visually impaired children.The registry is thus expected to have complete coverage of visually impaired children in Denmark.Registration is the gateway to multidisciplinary visual rehabilitation by ophthalmologists, optometrists, and low vision therapists.We used the register to analyse temporal changes in the prevalence and causes of childhood onset VI in the ex-preterm Danish population in the approximately 40 years that the registry covers.

| M ET HOD S
The study was a retrospective analysis based on registry data of PT children enrolled in the NRVIC at any time between 1988 and 2020.The register includes all Danish children with VI defined as visual acuity (VA) of ≤20/60, visual field defect ≤20 degrees or hemianopia.Patients can also be registered irrespectively of their visual function if they have a degenerative retinal disease.Furthermore, patients in whom VI is suspected but where visual acuity cannot be determined with certainty, for example, due to poor cooperation, and children who are under diagnostic evaluation can also be registered if their visual function is below that expected for their age group.Children are deregistered if a VI is no longer suspected, vision improves to a VA >20/60 or when they reach 18 years of age.
Children are enrolled in the NRVIC by a health care professional filling out an enrolment formula stating patient and parental information, cause of VI and VA.Most children (>97%) are enrolled by paediatric ophthalmologists (Paris-Horan et al., 2022).Upon enrolment, the NRVIC collects information from other healthcare professionals, for example, birth information, copies of medical files from paediatric departments or other relevant departments in addition to copies of medical files from the ophthalmic department that follows the child.After enrolment, the registry prospectively receives updated information from the treating physicians and from vision therapists.
The inclusion criteria for this study were children born prematurely and enrolled at NRVIC at any time between 1988 and 2020.Prematurity was defined as birth <32 of GA.In addition, patients had to survive the first year of life to be included.Exclusion criterion was deregistration before the age of 18 due to vision improvement that no longer justified enrolment in the register.Children registered with implausible birth parameters, for example, low gestational age and birthweight >2000 g were excluded.Children with an unknown status of visual function were also excluded.
The extracted variables for each patient were GA, birth weight (BW), sex, age at enrolment, birth year and severity of VI.Severity of VI was classified based on the visual acuity (VA) of the best-seeing eye.VI severity was sorted in mild VI (VA > 20/60), moderate VI (VA ≤ 20/60 to >20/200), severe VI (≤20/200 to >20/1250), and blindness (≤20/1250 to no light perception).
The presence of a severe visual field defect (hemianopia or visual field <20 degrees) reduced the VI severity by one group.Visual fields <10 degrees were characterized as severe VI.
In addition, we extracted information on the occurrence of other comorbidities from the NRVIC.The results were sorted in two groups: isolated VI and VI in addition to mental-, and/or physical impairment labelled 'combined impairments' in the rest of the manuscript.Lastly, to analyse temporal changes in causes of VI, the patients were subdivided by decade of birth: 1970s: 1970-1979, 1980s: 1980-1989, 1990s: 1990-1999, 2000s: 2000-2009, 2010s: 2010-2019.Medical information available in the NRVIC was reviewed to verify the cause of VI and update the terminology.A person may have had more than one condition impairing visual function but based on the judgement of two experienced paediatric ophthalmologists, we assigned the cause believed to have the greatest impact on visual function as the cause of VI.However, if an ocular cause of VI could be identified in a child with cerebral dysfunction, the ocular cause was chosen as the main cause of VI.Optic atrophy was used when the optic nerve was described as atrophic, pale, or white upon fundoscopy in the medical file.CVI was used when vision was impaired with symptoms characteristic of CVI (difficulties responding to visual stimuli and recognizing faces, or objects especially in cluttered spaces and difficulties understanding the visual input) in the absence of evident ocular disease, for example, optic atrophy, or retinal changes related to prematurity.

| Data access and management
We used the statistics program STATA SE 17 for data analysis.The data are presented in tables and figures using descriptive statistics as median values and inter quartile range (IQR) of 25-75.Prevalence of VI in children born PT, who survived first year of age, were obtained by comparing the information from the NRVIC to the publicly available information in the Danish Birth Registry (Sundhedsdatastyrelsen, 2022a(Sundhedsdatastyrelsen, , 2022b)).

| Approvals
As per Danish law (LBK number 1338, dated September 1, 2020), the study did not require approval from a medical ethical committee as we did not perform any patient examinations (decision number F-23066700 from the Committee on Health Research Ethics in the Capital Region).Access to information in the NRVIC was granted by institutional review.

| Prevalence of visual impairment in Denmark among preterm children
Three hundred and 61 (n = 361) ex-preterms had been registered in the NRVIC between 1988 and 2020.Fourteen (n = 14) were excluded from the study due to implausible GA for BW (e.g. a patient born at 20 GA and weighting 2900 grams), and 12 were excluded due to missing values (unknown VI severity n = 11 and comorbidities n = 1), resulting in 335 eligible patients.Due to a small number of patients (n = 3) born in 1970s these three patients were merged with the group of 1980s, resulting in: n = 76 from 1980s, n = 102 1990s, n = 83 2000s and n = 74 2010s.In the corresponding time period, a total of 19 013 children had been born prematurely.The prevalence of VI showed a decreasing tendency, affecting 26/1000 premature children born in 1980's to only 15/1000 in 2010s, respectively (Table 1).The prevalence of VI during the entire time period was 17/1000 child born preterm.

| Preterm children enrolled in the National Register of blind and visually impaired children
The majority of the enrolled patients were males (62%).Median BW was 1075 grams (g) (IQR 835:1480) and median GA was 28 weeks (w) (IQR 26:30).Median age at time of enrolment to NRVIC was 1 year (IQR 1:5) (Table 2).

| Main causes of visual impairment
Sequelae to ROP was the most common cause of VI followed, in decreasing order of magnitude, by optic atrophy, CVI, unknown causes, other retinal abnormalities, nystagmus, hemianopia, optic nerve hypoplasia, albinism and congenital cataract, see Table 3.The prevalence of ROP sequelae as the cause of VI decreased over the years from 51% in the 1980s to 34% in the 2010s, whereas CVI increased from 1% in the 1980s to 36% in the 2010s (Table 3).Sequelae to ROP included retinal detachments and other structural changes consistent with premature birth, as macular dragging, arrested macular development, retinal dysplasia, and nystagmus.ROP-related retinal detachment was present in 79 patients (24%) with the majority being born in the 1980s and 1990s but only in 11 patients since 2000.Twenty-six (8%) had been treated for ROP, of whom 24/26 received laser therapy and 2/26 received injection with Anti-Vascular Endothelial Growth Factor (Anti-VEGF).

| Severity of visual impairment
Seventy-five (22.4%) were blind, 53 (15.8%) had severe VI, 192 (57.3%) had moderate VI, and 15 (4.5%) had mild VI.The prevalence of blindness decreased from 10/1000, 5/1000, 3/1000 to 1/1000 in the 1980s, 1990s, 2000s and Note: The prevalence of visual impairment is reported per 1000 children born prematurely GA < 32, during the last four decades.a Data on number of premature born children were obtained from Sundhedsdatastyrelsen and includes only premature children who survived the first year of life (Sundhedsdatastyrelsen, 2022a(Sundhedsdatastyrelsen, , 2022b).Information about preterm birth divided by gestational age was not available before year 1996 (Sundhedsdatastyrelsen, 2022a).
2010s, (Figure 1).ROP-related retinal detachments were a common cause of blindness (n = 43/75, 57.3%).However, not all children with ROP-related retinal detachment were blind as some were only affected on one eye.Figure 2 demonstrates the temporal changes in cases with blindness sorted by the gestational age of the children.In the last decade (2010s), 6 children were registered as blind, five of them had CVI as the main diagnosis and the last one had received laser treatment for ROP.All 6 had combined impairments including VI.
or Anti-VEGF ROP was introduced in 2011 in Denmark.In this study only two patients received intra ocular injection with Lucentis and both did not develop retinal detachment, accordingly it is not possible to discuss a potential effect of this treatment on the prevalence of retinal detachment.A great proportion of children had systemic comorbidities in addition to their VI suggesting that the children and their families face a complex life after premature birth.A couple of Danish studies have previously described the main causes of VI prior to the national overview that was established with the NRVIC (Bech et al., 1993;Fledelius & Greisen, 1993;Rosenberg, 1987).The first manuscript describing low vision due to ROP was conducted in 1987, and presented data from 1948-1985, hence the period analysed in this study expands from 1988-2020.Based on these previous studies, severe VI due to ROP peaked in mid 1980s probably due to unfavourable outcomes in cases with ROP screening failures and lack of treatment, emphasizing the crucial importance of diagnosing and treating ROP according to current international guidelines (Chiang et al., 2021;Good, 2004;Mintz-Hittner et al., 2011;Palmer, 1990).A significant increase in the incidence of ROP treatment and decline in screening failures in Denmark was reported between 1996 and 2005 (Slidsborg et al., 2008).This supports the findings in this study, where we found a declining rate of VI due to ROP the last decade.The low prevalence of severe VI due to ROP in Denmark is in accordance with socio-economically comparable countries as Sweden, the Netherlands, and New Zealand (Norman et al., 2019;Tan et al., 2015;Van Sorge et al., 2011).
The effects of prematurity and ROP on the eye has been extensively studied (Fielder et al., 2015;Fieß et al., 2017;Fledelius et al., 2015;Pétursdóttir et al., 2021).The development in ROP treatment from cryotherapy in 1980s, laser in 1990s and to anti-VEGF in 2010s have had an impact on the ocular findings in preterm children who have undergone treatment for ROP (Good, 2004;Mintz-Hittner, 2012;Palmer, 1990).Previously, retinal detachment, pathological myopia and visual field defects were frequently reported after CRYO and laser therapy (Liang, 2019;Palmer et al., 2005).Today, the

GA≤27
characteristics of have changed with the increasing survival rate of extreme PT children.Anti-may be preferred as a treatment in cases with very aggressive posterior disease to preserve the visual field of the peripheral retina (Chiang et al., 2021;Mintz-Hittner, 2012).However, some concern exists regarding the impact of anti-VEGF treatment on the neurodevelopmental delay (Tan et al., 2021).
Ophthalmic and cerebral sequelae to prematurity may often be present in the same visually impaired patient.There seems to be an association between CVI and ocular complications in preterm children, both with and without ROP (Diggikar et al., 2023;Dutton, 2013;Jacobson et al., 2021).The most recent Danish study investigating the primary risk factor for VI in extreme PT was conducted in 2012 and included an examination of a cohort of children at 4 years of age (Slidsborg et al., 2012).The study found that cerebral damage was a more frequent cause of VI than ROP sequelae based on clinical examinations, fundus photographies and questionnaires.The study, however, was performed before the introduction of optical coherence tomography (OCT) in routine clinical practice.The present study showed an increase in the prevalence of CVI over the years, and that CVI was more commonly associated with moderate or mild VI.This is in line with other studies showing that preventable causes of blindness have decreased such as ROP and cataract in high-income countries, whereas CVI and optic atrophy has increased (Boonstra et al., 2012;Li et al., 2023).The definition and description of CVI was rather vague in most of the records reviewed for this study and awareness of the condition may have changed over the last four decades.Thus, some CVI cases might have been unreported in the 1980s and early 1990s.It seems advisable to arise awareness of the condition, especially among healthcare professionals managing ex-preterm children.
Additionally, the modern imaging systems can improve our diagnostic skills regarding optic atrophy.Rather than evaluating optic atrophy solely by fundoscopy supplementing the clinical examination, with OCT scans of the optic nerve and macula, can reveal thinning in the retinal nerve fibre and ganglion cell layer.In children, optic atrophy can occur due to damage to the anterior or posterior visual pathways (Scott et al., 2023).Further, preterm children may be exposed to some degree of asphyxia and brain damage, which both can lead to optic atrophy (Denne et al., 2003).
We expected to find an increased prevalence of combined comorbidities in children born GA ≤ 27 compared to GA 28-31 (Marlow et al., 2005;Pascal et al., 2018).Much to our surprise, we found that the very preterm had a higher prevalence of systemic comorbidities than the extremely preterm.An explanation to this might be, that some neurodevelopmental delays may not present until later in life and may not have been included in the regular updates of the registry if they were not reported in the ophthalmic part of the medical file (Battajon et al., 2023;Kaul et al., 2021).

| Strengths and limitations
The main strength of this study is that it is based on nation-wide data captured continuously over four decades, including information about ophthalmological and systemic comorbidities, allowing an analysis of the main reasons for VI among Danish preterms.
In this study we have not examined the maternal characteristics that can influence the characteristics of the preterms.A recent Danish study has shown that the risk of preterm birth is increased with decreasing maternal educational level and increasing mental health conditions (Knudsen et al., 2021).Further, we used a retrospective study design with data from the entire country, which means that many different ophthalmologists have seen the children and the terminology and diagnostic methods have changed over the years, for example, with the introduction of OCT in clinical practice.However, most children are enrolled after having been seen by an ophthalmologist specially trained in paediatric ophthalmology and generally the data quality in NRVIC is believed to be high (Paris-Horan et al., 2022).
VI severity for each patient was based on the best-seeing eye; however, the type of ROP sequelae was reported for the most unfavourable outcome in either eye for each patient, for example, the patient might have moderate VI but monocular retinal detachment due to ROP.This might have underestimated the number of favourable outcomes.Yet, we believe that this study presents unique results based on Danish registries that has proven valid and of high quality.

| CONCLUSION
The improvement of the standard neonatal intensive care units and management of ROP has reduced the incidence severe VI due to over the last in Denmark and blindness has been almost completely eliminated.However, many ex-preterm children with VI are also affected by systemic comorbidities and sequelae to premature delivery.The combined effects of multiple impairments may have a complex impact on children's physical, cognitive and social development that might affect them into adulthood.Further studies are needed to investigate the impact of systemic comorbidities on the visual function in preterm children.

F
Prevalence of visual impairment severities.The figure illustrates the changes in the prevalence of visual impairment (VI) severity in the approximately 40 years the Danish National Registry of Visually Impaired children covers.The presented VI severities: "Mild VI" group was only introduced in the 2000s (green line), Moderate VI (orange line), Severe VI (grey line) and lastly the Blind (red line).
Cases of blindness among preterm.The figure illustrates the changes in the cases of blindness sorted in gestational age (GA) in the approximately 40 years the Danish National Registry of Visually Impaired children covers.GA less than or equal to 27 weeks (red line), GA between 28-31 (green line) and lastly the total of both GA groups (blue line).
Prevalence of visual impairment per 1000 children born <32 GA in birth decades.
T A B L E 1 Characteristics of preterm children <32 GA enrolled in NRVIC.The table presents descriptive variables for the preterms enrolled in NRVIC sorted in birth decades.Descriptive results presented in median (IQR 25:75).
T A B L E 2Note: Comorbidities tabulated with cause of visual impairment.Combined impairments include mental and/or physical impairments beyond the visual impairment.
T A B L E 4Note: