Visual impairment and cognitive performance: A nationwide study of 1.4 million adolescents

Previous research highlights the adverse effects of visual impairment (VI) on academic achievement in children, yet its impact on cognitive performance among adolescents and young adults remains under‐studied. Therefore, this investigation aimed to analyse this association in a nationwide sample of Israeli adolescents.


INTRODUC TION
2][3][4][5][6] In contrast, studies within the elderly population have explored the links between deteriorating vision and cognitive decline or dementia. 7,8owever, there is a limited body of work examining the relationship between vision and cognition among adolescents or young adults.Given that in these age groups many cognitive functions are nearing their developmental peak, 9 and that preventable childhood ocular conditions such as amblyopia are less remediable, 10 such studies are important as they may fill a gap in the literature and reveal whether the negative vision-cognition association, previously identified in early childhood, is also evident during adolescence.This is also of value considering that early-life cognitive performance is linked with long-term sequelae including increased morbidity and mortality rates 11,12 and socio-economic success. 13he unique setting in Israel includes mandatory military service preceded by encompassing pre-recruitment medical and psychosocial examinations.We hypothesised that VI in adolescents is associated with decreased cognitive performance scores.Therefore, this population-based study, the largest of its kind to date, aimed to test this hypothesis and investigate the association between VI and cognitive performance scores in a sample of more than 1.4 million adolescents.

Study population
One year before mandatory military service, Israeli adolescents around 17 years of age undergo a pre-draft screening aimed at assessing their service eligibility.This process includes the collection of sociodemographic data and medical background, a thorough medical examination conducted by a physician, as well as cognitive and behavioural assessments.Best-corrected visual acuity (BCVA) and refractive error are also measured and documented. 14The study population included Israeli-born adolescents aged 16-19 years, who were examined from 1993 to 2017.Visual acuity data were not systemically documented before 1993, and complete annual data are currently accessible up to 2017.Individuals were included in the analysis if BCVA data were available for both eyes and cognitive assessment results were recorded.Ultra-orthodox Jewish adolescents, comprising 2.9% of the target age group, were excluded due to their infrequent evaluation and distinct educational system, which has been associated with an increased prevalence of overall myopia and especially high myopia. 15In addition, adolescents from non-Jewish minority populations were less represented in this cohort as they were not obliged to perform military service and were not systematically evaluated.The institutional review board of the Israel Defense Forces Medical Corps approved the study and participants' confidentiality was strictly maintained during data processing.Participants' informed consent was waived as the data were de-identified.The study adhered to the tenets of the Declaration of Helsinki for research involving human subjects.

Assessment of cognitive performance
Cognitive performance was assessed by the General Intelligence Score (GIS), which demonstrates a correlation >85% with the Wechsler Adult Intelligence Scale and is comparable to the Intelligence Quotient (IQ). 11,16,17This score has been used previously in epidemiological studies within the same cohort, 11 exhibiting robust external validity to other populations. 12The Wechsler tests have been widely used in individuals with vision reduction, with relatively high levels of assessors' satisfaction rates. 18xaminees were instructed to bring their prescribed optical correction devices, if applicable, for the duration of the assessment.The GIS assessment comprises four subtests: the Otis-R, which measures the ability to understand and perform verbal instructions; Similarities-R, which assesses verbal abstraction and categorisation; Arithmetic-R, which measures mathematical reasoning, concentration and concept manipulation and Raven's Progressive Matrices-R, which measures nonverbal abstract reasoning and visual-spatial problem-solving abilities (Figure 1; additional sample questions are available in Appendix S1).The complexity of tasks within each subtest gradually increases, and examinees are required to complete each subtest within a pre-specified time limit.Examinees' scores were integrated to create a validated measure of general

Visual acuity testing and VI definition
During pre-recruitment screening, adolescents underwent visual acuity and refraction assessments.These were carried out using a standard Snellen chart at a distance of 6 m, with each eye tested individually by a certified optometrist or ophthalmologist.Initially, unaided visual acuity was determined for each adolescent using a Snellen chart.Adolescents presenting with unaided visual acuity worse than 6/6-allowing for up to one letter error on the 6/6 line for either eye-then underwent noncycloplegic refraction.This step was performed using an automated refractometer (models included Nikon Speedy K, Nikon.com; Topcon KR-8000, KR-7000S and earlier versions, Topcon.com).Following automated refraction, a subjective refraction was conducted for verification purposes, including measurement of the BCVA in each eye.Adolescents capable of reading all but one letter on the 6/6 line without optical correction were presumed to have a refractive error of 0 dioptres. 14,15Unilateral VI was defined as BCVA worse than 6/9 in one eye only.Bilateral VI was defined as BCVA worse than 6/9 in both eyes.
Adolescents with BCVA equal to or better than 6/9 in each eye were considered to have normal visual acuity.A diagnosis of amblyopia was assigned to adolescents with reduced BCVA, in the absence of other causative ocular findings, based on medical records and clinical evaluations. 19Amblyopia-related VI was classified as unilateral or bilateral using the same BCVA thresholds.

Covariates
Sociodemographic variables were collected for each individual. 20Years of education were divided into complete (≥11 school years, the maximum possible at the time of evaluation) or incomplete (<11 years).Socio-economic status (SES) was based on the classification method of the Central Bureau of Statistics, determined by the site of residence and categorised into three groups: low, medium and high.Country of origin was determined by the birth country of the father, or the grandfather if the father was born in Israel, and was divided into five categories: former Union of Soviet Socialist Republics (USSR), Asia (non-USSR), Africa (excluding South Africa), Western (comprised of non-USSR Europe, America, South Africa, Australia and New Zealand) and Israel.Anthropometric measurements included weight and height, measured barefoot and in light clothing by certified medics.Body mass index (BMI) was calculated as weight (kg) divided by the squared height (m 2 ).BMI was classified according to the age-and sex-adjusted growth charts of the United States Centers for Disease Control and Sample question in the Raven's Progressive Matrices-R subtest.Examinees were required to assess the relationship between two given shapes, replicate the same pattern on a different shape and select the response in which the relationship between the two shapes was preserved.In this particular case, the top base of the given shape (cylinder) was removed, and the remaining shape was inverted.The correct answer was therefore (b).
Prevention, as this categorisation was validated for the Israeli population. 21

Statistical analysis
Categorical variables are presented as numbers and percentages, and continuous variables as means and SD.The distribution of baseline characteristics across the three cognitive Z-score groups was calculated.Multinomial logistic regression models were applied to estimate the odds ratios (ORs) and 95% confidence intervals (CIs) for low and high cognitive Z-scores across the BCVA groups.

Subgroup and sensitivity analyses
Several pre-specified subgroup and sensitivity analyses were performed, all utilising the fully adjusted models: 1.A sex-stratified analysis to account for an effect modification of the VI-cognitive performance relationship by sex. 2. To minimise confounding by coexisting illness, which may be associated with both VI and cognitive performance, the analysis was limited to adolescents with unimpaired health status based on the pre-recruitment medical evaluation.Unimpaired health was defined as the lack of any comorbidity that required chronic medical therapy or follow-up, including central nervous system insult or neurodevelopmental disorders affecting the visual system, or any history of cancer or major surgery. 20. To minimise the potential effect of VI on task comprehension during the cognitive test administration, we focused only on adolescents with mild VI, defined as BCVA worse than 6/9 in either or both eyes, and equal to or better than 6/18 in both eyes. 22Adolescents with BCVA worse than 6/18 in either or both eyes were excluded from this analysis.4. The analysis was limited to adolescents of medium SES to establish a more uniform socio-economic background, thereby minimising the confounding effect of low or high SES determinants on both VI and cognitive performance.
5. The analysis was limited to adolescents who had completed ≥11 years of education at the time of evaluation, aiming to reduce the confounding effect of low educational attainment on the association between VI and cognitive performance.6.To account for a potential time period bias, only adolescents who were examined between 2008 and 2017 (the last 10 years of the study period) were included.7. The analysis was limited to adolescents with amblyopia as a pre-specified cause for vision reduction.Adolescents with reduced BCVA attributed to other ocular conditions were excluded from this analysis.8. Since higher cognitive performance has been linked with increased myopia prevalence in a similar source population, 16 the analysis was limited to adolescents without myopia.Myopia was defined based on the right eye spherical equivalent (SEQ; calculated as sphere +0.5 × cylinder) ≤ −0.75 D. 9. A stricter threshold for the definition of VI was used, defined as BCVA worse than 6/12 in either or both eyes.

VI and cognitive performance
Figure 3 presents the fully adjusted ORs for low and high cognitive Z-scores per BCVA group among visually impaired and amblyopic adolescents.Overall, the unadjusted and partially adjusted models were consistent with the fully adjusted model (Figure 1a).In the fully adjusted model, adolescents with unilateral VI had increased OR for low cognitive Z-scores (1.24, 95% CI 1.19-1.30)and decreased OR for high cognitive Z-scores (0.84, 95% CI 0.80-0.88).Notably, ORs for low and high cognitive Z-scores were accentuated among adolescents with bilateral VI (1.62, 95% CI 1.50-1.75and 0.81, 95% CI 0.73-0.90,respectively).
Applying the fully adjusted model for each cognitive assessment component separately revealed comparable results, with the greatest effect size noted considering the visual-spatial functioning subtest (Figure 4).

Sensitivity and subgroup analyses
The association between VI and cognitive performance persisted in various sensitivity and subgroup analyses (Tables S1 and S2).The association was evident across multiple subgroups: adolescents of both sexes, those with unimpaired health (n = 956,723), medium SES (n = 339,431), complete years of education (n = 1,329,368) and without myopia (n = 1,009,528).We also found similar estimates of association in sensitivity analyses focusing on adolescents with amblyopia-related VI (n = 1,398,122, Figure 3), mild VI (n = 1,405,411) and when employing a stricter BCVA threshold as a definition of VI (n = 1,410,616).

DISCUSSION
This population-based nationwide study of over 1.4 million adolescents demonstrated that VI was associated with lower cognitive performance scores, as assessed by the GIS and its components.This relationship was evident, not only among adolescents with bilateral VI but also in with unilateral and mild VI.ORs remained stable following gradual adjustment for multiple sociodemographic confounders and withstood several sensitivity and subgroup analyses.
3][24] Growing evidence supports the paradigm that visual deficits negatively affect academic achievement in children [1][2][3][4]6 ; however, the results are mixed, 5,25,26 and numerous methodological issues limit the ability to draw definitive conclusions from the existing literature.27 For instance, the definition of VI varies between studies and may reflect either better or worse ocular function.Likewise, tools of cognitive performance assessment differ, and nonstandardised, self-reported or surrogate measures of intelligence are often employed.2,28 Most studies have used relatively small sample sizes, ranging from dozens 1-3 to thousands 5,6,25 of participants, and population-based data, particularly among youth, are lacking. Th current findings extend current knowledge relying on a vast populationbased screening process and indicate that lower cognitive performance among visually impaired individuals may also become manifest during late adolescence.
Several investigations have addressed directly the impact of naturally occurring or simulated reduced vision on cognitive test performance, mostly in the adult population.While some exemplified null effects, 29,30 others have demonstrated that specific cognitive domains, particularly non-verbal skills, are more susceptible to low vision. 31Contrary to these mixed findings, the present study showed that VI was associated with reduced cognitive performance scores across all four examined cognitive domains, with the visual-spatial subtest displaying the strongest correlation with VI.These results are supported by evidence suggesting a shared visual form processing mechanism linking reading abilities and arithmetic skills, 32 as well as a connection between visual information processing and mathematical competence. 1,3,33Visual-spatial skills and arithmetic performance also interact with each other, as suggested by commensurate regions of brain activation in functional magnetic resonance imaging (fMRI) 34 and mutual genetic factors. 35lternatively, visually impaired adolescents might attain lower cognitive performance scores because of reading difficulties, thereby possibly failing to demonstrate their full cognitive potential in a time-limited test.Indeed, children with visual deficits demonstrate lower literacy scores and decreased reading proficiency relative to their peers. 2,28Amblyopic children also present with slow reading, which has been attributed to the 'crowding' phenomenon, poor eye fixation and an number of saccades.Recent research indicates that the presence of amblyopia itself, rather than the extent of visual deficit, is correlated with reading speed, both under binocular and fellow-eye reading conditions. 10The current data potentially corroborate this pattern, as a consistent relationship was observed between VI and reduced cognitive performance scores even among adolescents with only mild VI (BCVA equal to or better than 6/18), as well as in cases where visual acuity lay within the normal range in one eye (unilateral VI).
It should be acknowledged that all four cognitive subtests employed in this study entailed a certain level of reading (Appendix S1), which, in turn, limited our capacity to attribute the observed decrease in cognitive performance to factors independent of reading difficulties.This complexity was further echoed in a meta-analysis emphasising the heterogeneity of cognitive profiles in individuals with reading difficulties, as well as the bidirectional pathways linking cognitive abilities and reading development. 37Therefore, the observed decrease in cognitive performance scores could be argued to represent a broader outcome of intertwined influences of reading skills and cognitive abilities, within the context of VI.This hypothesis, however, requires confirmation through dedicated studies within the fields of educational psychology and neuropsychological assessment.While we not confirm the exact underlying mechanism of the observed link between visual acuity and cognitive performance, this study may hold significance for public health.The findings suggest the potential benefit of incorporating vision-adjusted time extension as a compensatory intervention to address the challenge of slow reading among visually impaired youth.Importantly, the findings also prompt the re-evaluation of formal vision thresholds regarding eligibility for academic accommodations, 10 especially given the decrease in cognitive performance scores seen even among adolescents with mild VI.
Regarding modifications to standardised cognitive assessments, various adapted versions utilise haptic modalities intended to address the needs of visually impaired children and adolescents adequately. 38That being said, any testing adaptations must first be validated to allow comparability with other standardised instruments.
This study has several limitations.First, the crosssectional study design prohibits inferring a causal relationship between VI and cognitive performance.Second, we lacked data on the allocation of partial credit for visual acuity measurements.However, this limitation is expected to introduce a non-differential misclassification bias.Third, the use of the Snellen chart might have led to inaccurate assessments of low vision levels.Our definition of VI, that is, identifying visual acuity worse than 6/9, likely mitigates this limitation.Fourth, the assumption that 6/6 vision precludes uncorrected refractive error may overlook the impact of significant hyperopia on sustained near-vision tasks, potentially leading to an underestimation of the observed associations in this subgroup.Fifth, our data set lacked temporal granularity to capture the exact timing of VI onset and the progression of vision deterioration.We also lacked data on potential confounders such as parental cognitive assessment and former academic achievement.Nevertheless, the results withstood The association of visual impairment and amblyopia with (a) low and (b) high cognitive performance in fully adjusted models (Adjusted for sex, birth year, body mass index, residential socio-economic status, level of education and country of origin).BCVA, best-corrected visual acuity.CI, confidence interval.OR, odds ratio.several sensitivity and subgroup analyses, including those restricted to adolescents with unimpaired health status, amblyopia-related VI and without myopia, collectively displaying parallel results.Sixth, by focusing on visual acuity measurements, we might have overlooked other vision assessment aspects like visual fields, contrast sensitivity and binocularity.3][24] Finally, as noted, we could not ascertain whether decreased cognitive performance scores observed in visually impaired adolescents indicated genuine cognitive deficits or rather an effect related to reading difficulties.As research indicates a multidimensional relationship between cognitive development and reading abilities, 37 these findings call for further longitudinal studies to better understand these interrelations among youth with VI.The strengths of this investigation include the reliance on a large national registry encompassing three the and screeningbased collection of sociodemographic and medical data including BCVA measurements and the battery cognitive tests yielding the GIS.
In conclusion, this highlights an association between VI and reduced cognitive performance scores in a nationwide sample of adolescents, as measured by the GIS assessment.Importantly, this association was consistently evident in cases of both unilateral and mild VI.Further research is essential for determining the exact causal pathways contributing to the observed relationship

F I G U R E 4
The association of visual impairment with (a) low and (b) high cognitive performance by different subtests in fully adjusted models (Adjusted for sex, birth year, body mass index, residential socio-economic status, level of education and country of origin).BCVA, best-corrected visual acuity.CI, confidence interval.OR, odds ratio.
T A B L E 1