Body mass index and visual impairment in Israeli adolescents: A nationwide study

Previous research on the association between body mass index (BMI) and visual impairment (VI) in youth has reported inconsistent findings. We aimed to investigate this association in a national cohort of Israeli adolescents.

type 2 diabetes mellitus, cardiovascular disease, and certain types of cancer. 2 Concurring with the global trend, an increase of up to 2.1 kg/ m 2 in mean BMI has been recorded in Israel as well, culminating in one in every six young adults falling into the overweight or obese categories. 3,4Markedly, this trend was documented in both men and women and across an array of socioeconomic and ethnic groups. 5sual impairment (VI) is also a crucial public health problem, particularly among young individuals, as reduced vision negatively impacts academic performance, socialization, and overall quality of life. 6,70][11] While the Israeli National Registry of the Blind consistently reports annual incidence rates for childhood blindness and severe VI, 12,13 specific prevalence metrics for VI of all severities within the Israeli paediatric population are not available.
The relationship between BMI and VI is well-established in adults, whereby elevated BMI poses a risk factor for the development and progression of various ocular conditions, including cataract, glaucoma, diabetic retinopathy, and age-related macular degeneration. 14Nevertheless, the association between BMI and VI in the paediatric and adolescent population remains elusive, owing to a different set of ocular conditions resulting in VI and limited population-based investigations exploring the link between the two. 11,15e current study aimed to fill this gap in the literature by utilizing a national database comprising over 1.6 million Israeli pre-recruits to investigate the association between BMI and VI and assess its magnitude.

| Participants
In Israel, mandatory recruitment into the Israeli Defence Forces (IDF) applies to the majority of Jewish citizens.The pre-draft screening process, usually at 17 years of age, includes various assessments to determine military fitness, such as sociodemographic and medical history evaluations, a comprehensive medical examination by a qualified physician, and assessments of cognitive and behavioural functions. 16st-corrected visual acuity (BCVA) and refraction are also measured and documented. 17,18If abnormal findings or positive medical history arise during the medical assessment, the individual is referred for further testing or consultation with a board-certified specialist.The protocol concludes when a definitive diagnosis is ascertained and its impact on service eligibility is determined.
The study population included Israeli adolescents aged 16-19 years, examined at national recruitment centers between 1993 and 2017.Visual acuity data were not systematically recorded before 1993 and complete data were available until 2017.Adolescents were analysed if BCVA data were available for both eyes and anthropometric measurements were recorded.Ultra-Orthodox adolescents were excluded due to irregular screening protocols and distinct lifestyle characteristics that have been associated with increased prevalence of myopia. 18Adolescents from non-Jewish minority populations were underrepresented in the cohort because of their lack of obligation for military service.
The IDF Medical Corps Institutional Review Board approved this study and waived the requirement for informed consent based on strict protection of the anonymity of the individuals included.

| Examinations and definitions
Body measurements included weight and height measured barefoot and in underwear, using a beam balance and stadiometer. 16,19Body mass index (BMI) was calculated as the weight (kg) divided by the squared height (m 2 ).BMI was classified according to the age-(in months) and sexadjusted percentiles of the United States Centers for Disease Control and Prevention (CDC), which were validated for Israeli adolescents. 20BMI percentiles were grouped as follows: BMI < 5th (underweight), 5th ≤ BMI < 50th (low-normal), 50th ≤ BMI < 85th (high-normal), 85th ≤ BMI < 95th (overweight), mild obesity (class 1 obesity; ≥95th percentile to <120% of the 95th percentile), and severe obesity (class 2 and 3 obesity; any of ≥120% of the 95th percentile or BMI ≥35 kg/m 2 ).
BCVA was assessed separately for each eye using a standard Snellen chart at a 6-meter distance.Non-cycloplegic refraction measurements were obtained using an auto-refractometer (Speedy K; Nikon Corp., Tokyo, Japan; KR-8000 and KR7000S, Topcon, Tokyo, Japan; and earlier models).10][11]

| Study variables
Years of formal education were dichotomized as ≥11 or <11 years, with the assessment of the majority of adolescents during their 11th year of schooling.Residential socioeconomic status (SES) was derived from the Central Bureau of Statistics classification system, determined by residential locality, and categorized into three groups: low (1-4), medium (5-7), and high (8-10), (Supplementary Information). 5,16Cognitive function was assessed using a validated general intelligence test comparable to the intelligence quotient (IQ), with cognitive scores recorded on a 1-9 scale, and divided into three groups: low (1-3), medium (4-7), and high (8, 9), (Supplementary Information). 21The country of origin was determined by the father's country of birth, or the grandfather's if the father was born in Israel, and was divided into five areas: former Union of Soviet Socialist Republics (USSR) countries, Asia (non-USSR), Africa (excluding South Africa), Western (comprised of non-USSR Europe, America, South Africa, Australia, and New Zealand), and Israel.The country of birth was dichotomized as Israel or non-Israel based on data from the Israeli Ministry of Inferior.Adolescents were classified as having unimpaired health when no chronic morbidities were present, including central nervous system insult or neurodevelopmental disorders affecting the visual system, nor was there a requirement for ongoing medical treatment, or a history of cancer or major surgery. 16

| Statistical analysis
Descriptive statistics were calculated and presented for adolescents with normal vision and unilateral or bilateral VI.Continuous variables were described as means and standard deviations (SD) and categorical variables as numbers and proportions.One-way analysis of variance (ANOVA) test was used to compare continuous variables, and the chi-square test was used to compare categorical variables.BMI was investigated both as a categorical and a continuous variable.We tested sociodemographic variables potentially associated with VI based on the existing literature using univariable multinomial regression models.We then employed multivariable multinomial regression models with BMI as an independent variable and VI as an outcome variable, controlling for significant covariates found in univariable models ( p < 0.05).Normal vision and high-normal BMI categories served as reference groups.The multivariable models were stratified by sex to account for a hypothesized effect modification between BMI, sex, and VI.To corroborate this hypothesis, an interaction term was included in a binary logistic model comparing normal vision to overall VI (encompassing both unilateral and bilateral cases).Multicollinearity diagnostic tests were conducted, and the variance inflation factor (VIF) was <1.5 for all included variables.Finally, adjusted restricted cubic splines with knots at the 5th, 50th, 85th and 95th percentiles were used to model the dose-response relationship between BMI and VI, consistent with the cut-offs in the CDC-based BMI percentile categorization.The results of the models were reported as odds ratios (OR) with 95% confidence intervals (95% CI).A two-sided p-value <0.05 was considered statistically significant.Statistical analyses were performed with SPSS version 29.0 (IBM Corp., Armonk, NY, USA) except for spline models, which were generated with R version 4.0.2(R Core Team, Vienna, Austria).
We performed the following pre-specified sex-stratified sensitivity and subgroup analyses, all utilizing the adjusted model:  c.Body height (in centimetres) was added as an additional covariate in the main analysis to control for its potential effect on ocular metrics, aside from its contribution to BMI. 22 d.We limited the outcome variable to amblyopia-related VI.In this analysis, adolescents with VI from other causes were excluded.
The definition of amblyopia used for this purpose is described in the Supplementary Information.
e. We limited the outcome variable to mild VI, defined as BCVA worse than 6/9 and equal to or better than 6/18 in both eyes.In this analysis, adolescents with BCVA worse than 6/18 in either eye were excluded.
f.We analysed the association between a quadratic BMI term (BMI 2 /100) and VI.

| Cohort characteristics
Table 1 presents the baseline characteristics of the 957 169 men and 739 891 women included in this study.129 218 (7.1%) adolescents were excluded from data analysis based on missing visual or anthropometric data, or due to ultra-orthodox orientation (Figure 1).
The overall prevalence of unilateral and bilateral VI was 1.05% (n = 17 871) and 0.30% (n = 5148), respectively.Compared with adolescents with normal vision, those with VI were less likely to be of high socioeconomic status, obtain a high cognitive function score, achieve ≥11 years of education, be born in Israel, and be of Israeli origin.

| BMI and VI
There was a statistically significant interaction between BMI, sex, and VI ( p interaction <0.001).Figure 2 depicts the adjusted ORs and 95% CI for unilateral and bilateral VI across the entire BMI spectrum, based on the restricted cubic splines models.In both men and women, the models suggested a non-linear J-shaped relationship between BMI and VI, wherein the minimum ORs were observed within the range of normal BMI.

| Sensitivity and subgroup analyses
In sub-group analyses focusing on men with unimpaired health and those without moderate-to-severe myopia, ORs for both unilateral and bilateral VI were consistent with those in the main adjusted analysis, exhibiting a J-shaped pattern.This consistency was also maintained when the outcome variable was limited to amblyopia-related VI or mild VI, and when height was added as an additional covariate to the main analysis (Table S2).
Sub-group analyses of women with unimpaired health and those without moderate-to-severe myopia revealed consistent ORs for unilateral VI, but fluctuating ORs for bilateral VI, occasionally reaching statistical significance.This was also evident when the outcome variable was limited to amblyopia-related VI or mild VI, and when height was added as an additional covariate to the main analysis (Table S3).
The quadratic BMI term was significantly associated with unilateral and bilateral VI in both men and women (Table S4).

F I G U R E 1 Study population build-up.
In this large population-based study of over 1.6 million adolescents, BMI was associated with VI in a J-shaped pattern.High BMI was linked to increased odds for unilateral and bilateral VI in a dosedependent manner in both sexes, whereas low BMI was more markedly associated with VI in men.These findings were evident following adjustment for potential sociodemographic confounders and persisted in several sensitivity and subgroup analyses.
Previous research on the association between body measurements and VI in young populations is inconclusive, although most studies indicate a positive association between high BMI and VI. 11,23r instance, a study involving 3771 Chinese children and young adults aged 6-21 years found that BMI ≥19.81 kg/m 2 was connected to a higher VI prevalence. 24In contrast, a nationally representative study of 23 890 American adolescents utilizing parents' reports on their child's vision showed no association between overweight or obesity and VI. 25 Our study, leveraging a large population-based dataset, contributes to the existing literature by assessing the magnitude of the association between BMI and VI.While prior research has predominantly used collective definitions or non-objective measurements of VI, the present study provides greater granularity by examining VI laterality (unilateral vs. bilateral) and analysing a subset of adolescents with mild VI (BCVA ≥6/18).Importantly, BMI correlated even with a seemingly minor reduction in BCVA, indicating an association at a low threshold yet to be described.
VI can be differentiated by the specific pathologies responsible for reduced visual acuity.Evidence from paediatric research points to several ocular alterations linked with obesity, each potentially contributing to VI. Examples include denser crystalline lens, 26 higher intraocular pressure, 27 retinal microvascular abnormalities, 28 decrease in retinal nerve fibre layer thickness, 29 decrease in macular and subfoveal choroidal thickness, 30 and the development of increased idiopathic intracranial pressure. 31The exact pathophysiological pathways underlying these changes are not fully recognized, but obesity-related factors such as hyperleptinemia, induced inflammatory response, endothelial dysfunction, and oxidative stress are thought to play a role. 14,32In this respect, a major caveat of this study is the lack of  14 This suggests that increased BMI might potentially induce changes leading to VI not exclusively mediated by these conditions.
Alternatively, the BMI-VI relationship might operate bidirectionally.Factors such as decreased physical activity, restricted access to healthy food options, and the psychological strain associated with impaired vision might contribute to BMI fluctuations, encompassing both underweight and obesity. 25,33,34The estimated effect size was generally larger among adolescents with bilateral vs. unilateral VI, supporting the complexity of the observed relationship.One plausible explanation for this pattern could be the more pronounced effects of bilateral VI on daily activities and quality of life, which in turn might amplify changes in BMI. 35nally, other common factors might influence both BMI and VI, thereby shaping their interconnection.Myopia, affecting up to 30% of our source population, 36 similarly shares a J-shaped association with BMI 19 and might culminate in ocular changes leading to BCVA reduction. 37In this context, an analysis of adolescents without moderate-to-severe myopia demonstrated comparable point estimates, thereby weakening the possibility of confounding by myopia as a cause for VI.In addition, low socioeconomic status could also affect the BMI-VI association by impeding access to healthcare services, including eye care, 38 and promoting obesogenic behaviours. 5This, nonetheless, is probably not the only explanation, as the identified associations proved steady when accounting for residential SES and other related sociodemographic characteristics such as education level, cognitive function, and country of birth.
Although the precise underlying mechanisms involved in the BMI-VI relationship remain undetermined, our findings bear relevance to public health.VI might lead to reduced academic achievement, decreased productivity, lower quality of life, and increased healthcare expenses. 6,7,39Suggesting an intricate association between BMI and VI, this study highlights the need for multi-faceted and coordinated strategies that address both systemic health and ocular health in This study has several limitations.First, owing to its crosssectional design, no causal inference or direction of effect could be concluded between BMI and VI.Instead, a bidirectional relationship should be considered, wherein BMI may affect VI, and vice versa.
Second, the study sample consisted of Israeli adolescents, potentially limiting the generalizability of the outcomes to broader populations.
However, the alignment of our findings with global BMI and VI prevalence data, 4,9-11 the analyses adjusted for various sociodemographic variables, and the genetic heterogeneity of the Jewish population 41 lend support to this study's external validity.Third, we lacked data on possible confounding variables like dietary intake, outdoor exposure, and physical activity patterns, which restricted our capacity to assess their effect on the observed associations.Nevertheless, our results remained stable even among adolescents with unimpaired health, among those without moderate-to-severe myopia, and among those with amblyopia-related VI, demonstrating the robustness of our findings.Fourth, we recognize the unavailability of other adiposity indices, such as waist circumference and skinfold thickness, which could potentially provide a more accurate understanding of the investigated associations. 42Still, The U.S. Preventive Services Task Force recommends BMI as the preferred screening tool for obesity. 43Fifth, as a. To account for possible confounding by other comorbidities, we included only adolescents with unimpaired health status, as evaluated during the medical screening process.b.In light of recent evidence indicating an association between BMI and myopia using a similar dataset, we excluded adolescents with moderate-to-severe myopia (defined as right eye spherical equivalent ≤ À3.0 diopters) to address the possibility of confounding by BCVA reduction associated with myopia.19 young populations.This could include interventions that promote physical activity and dietary modifications specifically in young individuals with VI,23,25,40 along with the prompt identification and management of ocular disorders in those with abnormal BMI.Such efforts may help mitigate the societal and economic burdens related to these conditions.

F I G U R E 2
Odds ratios and 95% CI for unilateral and bilateral visual impairment by body mass index (BMI) in (A) men and (B) women.Restricted cubic spline models were fitted using knots at the 5th, 50th, 85th, and 95th BMI percentiles and adjusted for residential socioeconomic status, cognitive function score, years of education, country of birth, country of origin, and year of birth.The horizontal grey dashed line indicates an odds ratio of 1.The vertical red dashed lines correspond to BMI values associated with the minimum odds of visual impairment.mentioned, while our primary outcome of VI was based on a conventional definition using BCVA measurements, we did not delve into specific visual conditions that may lead to VI, limiting the understanding of the associations across different ocular pathologies.Future research should investigate these specific conditions, uncovering possible unique mechanisms linking BMI and various causes of VI.The strengths of this study lie in its methodical collection of sociodemographic, BMI, and BCVA data for a uniform age group of both sexes, management of potential sociodemographic confounders, and the analysis of a large sample size that enabled population-level examination spanning three decades.In conclusion, our results revealed a J-shaped association between BMI and VI in late adolescence, observed even in adolescents with unilateral VI, those with mild VI, and those who are otherwise healthy.While establishing a cause-and-effect relationship extends beyond the scope of this study, it underscores the importance of an integrated health approach that concurrently addresses BMI and VI in adolescents.Future investigations in diverse populations are essential for a nuanced understanding of the BMI-VI relationship, focusing particularly on distinct etiologies of VI.

1
Baseline characteristics of the study cohort.
Odds ratios for unilateral and bilateral visual impairment by body mass index.Adjusted for year of birth, residential socioeconomic status, cognitive function score, years of education, country of birth, and country of origin.
T A B L E 2 Abbreviations: BMI, body mass index; CI, confidence interval; OR, odds ratio; VI, visual impairment.a