5
Onset and Progression of Myopia

This chapter covers genetic and environmental factors that may contribute to the onset and progression of myopia. Evidence is presented on how myopia development may be influenced by genetic factors, visual behaviors such as near work or use of electronic devices, and the visual environment while indoors versus outdoors. Here the committee defines “myopia onset” as the first occurrence of a refractive error that meets the formal definition of myopia, usually a spherical or spherical equivalent refractive error between –0.5 and –1.0 D.¹ While most of this chapter focuses on human studies and randomized clinical trials, results from animal studies are also included as evidence for, or against, the effects of particular risk factors. The committee had the opportunity to develop recommendations from a holistic, societal perspective to address what is controllable about environmental risks for myopia, management of the condition, and improving equity in access to diagnosis and treatment of myopia.

During the past decade, insufficient time spent outdoors has emerged as a major factor for increasing the risk of myopia onset. The risk of myopia caused by insufficient time outdoors is amenable to study in randomized controlled trials, which have yielded convincing evidence for its role in myopia onset and, potentially, myopia progression. Researchers have also explored the possible roles of education and near-work, including the use of electronic devices such as smartphones. These investigations have relied predominantly on cross-sectional and longitudinal epidemiology study designs, which require strong assumptions for causal inference. Existing research suggests that genetics does not explain the surge in myopia prevalence over recent decades. Nevertheless, genetics research has provided insight into causal biological mechanisms and helped to explain differences between individuals in their susceptibility to environmental risk factors. Much of the change in myopia incidence is likely attributable to changes in exposure to lifestyle-related risk factors, the potential interaction of lifestyle and genetic risk factors, and the ways that parents and society manage children’s experiences. See Box 5-1 for an explanation of the correlation between earlier-onset myopia and higher amounts of myopia.

___________________

¹ As a reminder, refractive error covers the spectrum from farsightedness (hyperopia) to nearsightedness (myopia), measured in units called diopters (D).

GENETIC FACTORS ASSOCIATED WITH MYOPIA

The role of genetics in the etiology of refractive errors has been of long-standing interest, and its importance has been demonstrated by twin, family, and gene-level association and linkage analysis studies (Baird et al., 2020; Tedja et al., 2019). Modern lifestyles in high- and middle-income countries expose the majority of the population to risk factors for myopia (He et al., 2015; Mountjoy et al., 2018). Thus, an often-misinterpreted consequence of this ubiquitous exposure to lifestyle risk factors is that genetic differences explain much of the variation in refractive error between individuals, despite genetics not being the primary ‘driver’ of myopia development. In general, genetic differences between individuals confer upon each person a relatively high or low level of susceptibility to myopia when they are exposed to environmental risk factors, rather than acting in a purely deterministic manner (Pozarickij et al., 2019).

Genes Associated with Myopia and Refractive Error Development

The variance in a trait explained by genetic differences between individuals is termed heritability. The heritability of refractive error varies from population to population but is typically within the range of 30–80% (Sanfilippo et al., 2010). Genome-wide association studies (GWAS) have been extremely successful in identifying specific genetic variants associated with susceptibility to myopia (Tedja et al., 2019). A powerful approach has been to meta-analyze multiple datasets (mostly population-based) through collaborative consortia such as CREAM (the Consortium for Refractive Error and Myopia). In the most recent GWAS meta-analysis, which assessed a total of more than half a million participants, 449 genomic regions harboring refractive error-associated genetic variants were identified (Hysi et al., 2020). GWAS findings

have been very highly reproducible both within and across European ancestry and Asian ancestry study samples. African ancestry GWAS samples have been underrepresented to date.

Current evidence suggests that most genetic risk variants associated with myopia are shared between European and Asian ancestry groups and have similar effect sizes, which despite differences in allele frequencies between ancestry groups argues against genetics as an explanation for the widely differing prevalence levels observed between these geographic regions (Tedja et al., 2018). More research on mechanisms is needed to better understand how genetic variants affect refractive error development.

A current gap in knowledge is the mechanism by which the hundreds of known refractive error-associated variants confer susceptibility to myopia. (See Box 5-2 for information about mechanisms regulating normal and myopic eye growth.) Existing research to address this question has taken advantage of the genetic tractability of mouse and zebrafish models (Koli et al., 2021; Mazade et al., 2024; Quint et al., 2023; van der Sande et al., 2022). However, the difficulty of detecting subtle phenotypic effects in mice and zebrafish is a limitation, since most genetic variants discovered in GWAS analyses have only small effects. In addition, most GWAS variants are in non-coding regions of the genome, suggesting their effects are mediated not by the elimination of the protein product of a gene, but rather by altering a nearby gene’s level of expression or the timing and context of its expression. Completely knocking out a gene in a mouse or zebrafish is arguably a poor model for studying this type of genetic variant, even if such models can provide insight into the key biological pathways.

The term ‘monogenic’ refers to a genetic disorder caused by a mutation in a single gene. Monogenic disorders often affect several members of a family, due to the segregation of the mutation through the pedigree. The causative mutation often varies from one family to another, and a mutation in a different gene can give rise to the same disease. Mutations in approximately 20 different genes have been identified as the cause of non-syndromic, monogenic high myopia (Cai et al., 2019). High myopia is more likely to be monogenic in origin if it has an onset prior to the age of 7 years. For instance, in a study of 298 individuals with early-onset high myopia, rare mutations in nine genes in nine individuals (3% of the cohort) were identified with high probability as being the cause of the high myopia (Jiang et al., 2015). Mutations that give rise to monogenic high myopia usually have a direct, adverse functional impact and thus may offer insight into the etiology of myopia in the general population; alternatively, they may simply identify genes and proteins that need to be functional for the eye and visual system to develop normally (see Box 5-2).

A current gap in knowledge is the contribution of rare genetic variants to refractive error in the general population and to the burden of high myopia. Past studies have had limited statistical power to identify rare variants associated with high myopia (Guggenheim et al., 2022a; Su et al., 2023). Addressing this issue will require the assembly of very large cohorts of high myopia cases and controls, subjected to whole-genome sequencing.

Syndromic Myopia

A disease ‘syndrome’ refers to a cluster of clinical features that tend to co-occur; disease syndromes often have a monogenic origin. Myopia and high myopia are frequent features of monogenic disease syndromes (Flitcroft et al., 2023; Wildsoet, 1998). For example, characteristic clinical features of Stickler syndrome are high myopia and hearing loss (Richards et al., 2010). Each monogenic disease syndrome is generally rare, yet in aggregate syndromic myopia accounts for a significant proportion of cases. This is most evident for early-onset high myopia, where up to 40% of cases may be monogenic/syndromic in origin (Logan et al., 2004). A recent study by the Myopia Associated Genetics and Intervention Consortium examined 75 candidate genes in a sample of 6,215 school-aged children with high myopia (Yu et al., 2023). Putative disease-causing mutations were identified in 15% of the cohort. Preliminary research suggests there exists an above-chance level of overlap between genes that cause syndromic myopia and those implicated in ‘common’ myopia (Flitcroft et al., 2018).

Myopia is very often a feature of certain inherited retinal dystrophies, including congenital stationary night blindness and subtypes of retinitis pigmentosa (Hendriks et al., 2017). The mutations that cause these syndromes have been shown to cause myopia or increased myopia susceptibility in mouse models (Mazade et al., 2024) and also have roles in ON-bipolar cell signaling, one of two main pathways in the retina that process visual information from photoreceptors to retinal ganglion cells (Hendriks et al., 2017). How these mutations lead to myopia is an area of active investigation (Zeitz et al., 2023). Apart from a genetic cause, early-onset myopia can also occur as a result of prematurity, especially in infants with retinopathy of prematurity (Mao et al., 2019).

The OPN1LW and OPN1MW genes encoding the cone opsin photopigments, most responsive to long-wavelength red light (L-opsin) and middle-wavelength green light (M-opsin), are located together on the X-chromosome; the high level of sequence similarity between these opsin genes predisposes them to mutations. A rare combination of sequence variants in the OPN1LW gene can cause Bornholm Eye Disease (BED), a disease syndrome with the clinical features of high myopia, color vision deficiency, reduced visual acuity, and reduced cone responses on electroretinography (ERG; McClements et al., 2013). A notable finding in BED is a reduction in the amount of opsin photopigment in affected L-cones, often associated with a molecular defect known as exon-skipping. This results in the retinal cone mosaic of patients with BED containing a mixture of normal M-cones and viable but poorly responsive L-cones. It has been hypothesized that this mixture of normal M-cones and abnormal L-cones promotes eye growth and myopia via spurious activation of ON- and OFF-bipolar cells when viewing low-contrast images (Neitz et al., 2022). While still speculative, this hypothesis has been extended by suggesting a role for commonly occurring L-opsin gene variants with mild effects, and the ratio of L vs. M-cones, in predisposing individuals to low myopia (Neitz et al., 2022). The hypothesis has also been suggested to relate to the mechanism of spectacle lenses designed to slow myopia progression by subtly lowering retinal contrast (marketed as SightGlass Vision DOT^® lenses)

(Rappon et al., 2023). This topic is discussed further in the section on Reduction in Contrast vs. Peripheral Myopic Blur as Mechanisms of Spectacle Treatments, in Chapter 7.

Gene–Environment Interactions

If the hypothesis is correct that genetic factors determine susceptibility to myopia while environmental risk factors trigger myopia development and progression (Tedja et al., 2019), then the majority of the hundreds of known refractive-error-associated genetic variants probably exert their effects through gene–environment interactions. Attempts to identify specific genetic variants with gene–environment interaction effects on refractive error in humans are generally in agreement with this hypothesis, yet these studies have been far less fruitful than GWAS analyses aiming to detect direct genetic effects (Clark et al., 2022; Pozarickii et al., 2019; see Box 5-3 on GWAS). The reasons for this lack of success and the poor record of replication for gene–environment interaction studies are twofold. First, biobank-scale genetic studies rarely measure myopia-related risk factors, such as time spent outdoors during childhood, which limits the available sample size for gene–environment interaction studies. Second, the statistical power to identify gene–environment interaction effects is necessarily lower than for detecting direct genetic effects. To date, only a handful of genetic variants with replicated gene–environment interaction effects have been discovered; in all cases, the variants interact with education level, such that the risk allele is associated with a greater shift in refractive error toward myopia as the level of education increases (Baird et al., 2020; Clark et al., 2022).

Gene–environment interactions have also been investigated with animal studies, particularly in mice, in which both genes and environment can be manipulated. Some of these experiments have revealed that the genetic defect alone was not sufficient to generate a myopic phenotype. However, combining myopigenic stimuli with the genetic defect demonstrated increased susceptibility for myopia. For instance, mice with mutations in the ON-pathway (nyx) (Pardue et al. 2008) or causing retinal degenerations (Pde6b^rd1 and Pde6b^rd10; Park et al., 2013) had normal hyperopic refractions throughout the juvenile ocular development period, yet these mice also showed increased susceptibility to experimental myopia using form deprivation (Mazade et al., 2024). Alternatively, a knock-out model of Aplp2, a gene associated with glycinergic amacrine cells that have an inhibitory role in the retina, resulted in hyperopia and reduced susceptibility to form-deprivation myopia (Tkatchenko et al., 2015). These animal studies support the hypothesis that the visual environment interacts with genetic factors to determine whether myopia develops and to what extent.

traits that do not cluster strongly in families; for rare diseases that do cluster in families, in-depth genetic evaluation of the individual families works better. Participants in a GWAS for a disease are selected based on their status as a disease case or as controls. Participants in a GWAS of a trait are typically chosen as a representative sample of the full population. To perform a GWAS, each and every common genetic variant in the human genome is tested, in turn, for an association with the disease status or trait. This step-by-step approach gives each genetic variant an equal chance of being discovered as a disease-associated risk factor; hence a GWAS is sometimes referred to as being ‘hypothesis free’. There are several million common genetic variants in the human genome, so a GWAS entails performing this many separate statistical tests.

To account for all this statistical testing, which is prone to false positive discoveries, geneticists have established a p-value threshold of P < 5 × 10^-8 for declaring genome-wide significant association. To have a realistic chance of reaching such a stringent p-value threshold, GWAS sample sizes need to be very large. To make matters worse, genetic predisposition for most common diseases and traits tends to be spread widely across thousands of genetic variants scattered throughout the genome (a so-called ‘polygenic’ genetic architecture). Thus, most genetic variants have a tiny impact on disease risk. This means that GWAS sample sizes of hundreds of thousands or even millions of participants are needed (see Figure 5-2).

Polygenic Scores (PGS) and Polygenic Risk Scores (PGRS)

A GWAS for refractive error yields a series of regression coefficients quantifying the shift in refractive error toward myopia or hyperopia associated with carrying one or two copies of the non-reference allele for thousands or millions of genotyped or imputed genetic variants from across the human genome. A polygenic score reverses the logic of a GWAS analysis. First, the shift in refractive error toward myopia or hyperopia is identified at each locus, corresponding

to the GWAS regression coefficient multiplied by the number of copies of the non-reference allele. This is done for each of the thousands or millions of genetic variants carried by an individual person. The resulting values are summed to produce a single value quantifying their genetically predicted refractive error (Sugrue & Desikan, 2019). Initial PGS studies took account of just the top GWAS variants with the strongest association to refractive error (Hysi et al., 2020; Tedja et al., 2018), while later studies demonstrated improved accuracy by incorporating up to a million genetic variants (Clark et al., 2023b; Kassam et al., 2022).

The accuracy of a PGS is quantified as the ‘incremental R-squared’ in an independent test sample. That is, accuracy is measured as the variance in refractive error explained by the PGS over and above that explained by demographic characteristics such as age and sex. Current PGS for refractive error have an incremental R-squared of approximately 20% (Clark et al., 2023b). To put this level of prediction accuracy in context, the main known environmental risk factor for myopia, time spent outdoors (quantified objectively using a spectacle-mounted light level sensor), explained only 3% of the variance in refractive error in a (non-independent) test sample (Li et al., 2020). (However, see Figure 5-3 for a discussion of why the predictive accuracy of environmental risk factors may be underestimated compared to genetic factors.)

Whereas PGS predicts the level of a quantitative trait, PGRS identify individuals who are at high risk of developing a specific disorder (Sugrue & Desikan, 2019). PGRS for high myopia can predict the condition with an area under the receiver operating characteristics curve of approximately 80% in independent samples (Clark et al., 2023b; Hysi et al., 2020). This level of prediction accuracy approaches that required for use in the clinic. Compared to existing methods of predicting high myopia development, PGRS have the unique advantage of being able to be implemented in young children before the gradual transition from hyperopia to myopia starts to occur. This makes them well-suited to identifying ‘pre-myopic’ children who would benefit from prophylactic treatment interventions such as atropine eye drops (Yam et al., 2023) and increased time outdoors. Importantly, current PGRS for high myopia perform far better in individuals of European ancestry compared to those of non-European ancestry (Clark et al., 2023b; Kassam et al., 2022). Thus, a current research priority is to narrow the performance gap of polygenic scores across ancestry groups (Kachuri et al., 2024).

PGS and PGRS of sufficient predictive power have the potential to improve the assessment of myopia interventions in two different ways. First, because by definition they account for relevant genetic variation within a population, PGS can serve as regressors that reduce standard errors and thus allow for the better discernment of treatment effects, particularly in the context of gene–environment interactions (e.g. Barcellos et al. 2018). Second, using PGS or PGRS to select participants in clinical trials can potentially improve their efficiency through a combination of “prognostic enrichment” (effectively increasing the event rate by selecting people at higher risk for the disease) and by “predictive enrichment” (effectively increasing the treatment’s effect size).

For example, in a retrospective analysis of trials of statins for cardiovascular disease, Fahed et al. (2022, p. 2) showed that:

While collecting such data might be prohibitive for any given clinical trial and would require careful ethical oversight, in an official statement the American Heart Association has pointed out the value of a single concerted national effort to conduct broad genetic profiling of a large population, combined with detailed health information, to aid in the treatment and diagnosis of cardiovascular disease (O’Sullivan et al., 2022). Such a project would also have real benefits for better understanding gene–environment interactions in myopia.

Genetic Contribution to Myopic Maculopathy and Retinal Detachment

Aside from its value in helping to identify children at an increased risk of developing high myopia when they get older, genetic profiling might also be useful in patients with existing high myopia to identify those at greatest risk of myopic maculopathy or retinal detachment—although whether such a risk assessment is possible remains a current gap in knowledge. If it is possible, this knowledge would be valuable in stratifying patients to receive more or less frequent routine clinical follow-up. Patients with high myopia due to Stickler syndrome are at a high risk of retinal detachment (Richards et al., 2010) and GWAS analyses of retinal detachment

cases and controls have observed an overlap with myopia-predisposing loci, as well as unique genetic risk factors (Johnston et al., 2016).

Genetic predisposition to myopia, as quantified by a polygenic score, was found to be no better at predicting the risk of myopic maculopathy than knowledge of the degree of high myopia (Sugrue & Desikan, 2019). To date, GWAS analyses of cases with high myopia and myopic maculopathy vs. highly myopic controls free from maculopathy have utilized small sample sizes, which have limited power to identify risk loci (Hosoda et al., 2018; Wong et al., 2019). Future research to address this question will require the assembly of a very large cohort of patients with high myopia, who have been genotyped and assessed clinically for signs of myopic maculopathy or retinal detachment.

Epigenetics

Epigenetic effects are defined as changes in gene expression (or other phenotypes) that occur without a change to the underlying DNA sequence. The most well-studied epigenetic mechanisms involve DNA methylation and histone modification, which are dynamic processes operating in the cell nucleus. Research into the role of epigenetics in myopia development has been limited, with just two large-scale genome-wide epigenetic studies reported to date. One study of 3-year-old infants with myopia vs. non-myopic controls identified five specific genomic regions that had a reduced level of methylation. The other study reported a link between myopia and the paternal grandmother’s smoking during pregnancy, which the authors argued implicated a trans-generational epigenetic mechanism (Seow et al., 2019; Williams et al., 2019b). Research investigating methylation levels in specific regions of the genome has also been carried out; for example, one recent study targeted regions harboring miRNA-encoding genes was associated with high myopia (Swierkowska et al., 2022).

Epigenetic studies may hold promise by providing insight into how lifestyle risk factors lead to myopia development. However, existing studies have been underpowered due to small sample sizes, lack of longitudinal measurement of methylation levels and refractive error, and potential bias from confounding factors. As an example of the potential for confounding, smoking is known to produce changes to the epigenome (Wiklund et al., 2019), so the negative association that also exists between smoking and education level would be expected to lead to differences in methylation status between myopia cases and controls.

Parental Myopia as an Indicator of Genetic Risk

Parental myopia is a well-established risk factor for myopia and, accordingly, has been included in some myopia prediction algorithms. Initial research suggests that parental myopia and a polygenic score for refractive error provide complementary predictive information, consistent with the premise that parents with myopia expose their children to a more myopigenic home environment (such as one where reading indoors is favored over playing or working outside) than non-myopic parents (Mojarrad et al., 2018). Simply counting whether a child has zero, one, or two parents with myopia ignores valuable information concerning the exact refractive error of parents; accordingly, the component of parental myopia corresponding to purely genetic effects (i.e., excluding the home environment component) has a theoretical upper limit of explaining about 5% of the variation in a child’s refractive error (Guggenheim et al., 2017). This compares poorly with current polygenic scores, which already can explain approximately 20% of the variation in a child’s refractive error and, in theory, have the potential to explain as much as 30% to 50% of the variation (Clark et al., 2023b).

It has recently been discovered that for some traits, genetic predisposition is not, for the most part, transmitted directly from parents to children via the germline, but instead is transmitted indirectly via the home environment that parents create for their children (a phenomenon termed ‘genetic nurture’; Guggenheim et al., 2017). Most notably, about three-quarters of a child’s genetic predisposition to educational attainment (heritability = 17%) is transmitted via genetic nurture, while one-quarter is transmitted directly via the germline. At present, the available evidence suggests that the transmission of genetic predisposition to refractive error occurs predominantly via direct transmission through the germline rather than via genetic nurture (Guggenheim et al., 2022).

ENVIRONMENTAL FACTORS

While genetic factors influence refractive development and the prevalence of myopia, genetic factors in a population do not change rapidly. Given the rapid increase in myopia prevalence which began some decades ago, environmental factors have been implicated as driving the “myopia boom” (Dolgin, 2015). This section highlights the potential influence on myopia of near work, visual environment, and behavior.

Near Work—Re-Examining a Classic Risk Factor for Myopia

Near work, meaning any activity requiring ocular accommodation for clear vision at a close working distance, has been considered a risk factor for myopia for centuries. Below, we detail the evidence for near work as a risk factor for the onset and progression of myopia and discuss the limitations of prior studies on the topic.

Near Work and Education

Research over several decades has documented a close link between education and myopia (Morgan & Rose, 2004). It has been noted for more than 150 years that the prevalence of myopia is higher among professionals and those engaged in “white collar” indoor jobs requiring more years of schooling compared to those whose livelihoods involve more time outdoors, such as manual laborers or workers in agriculture (Goldschmidt, 1968). Ware (1813) believed that the higher prevalence of myopia among officers in the military and among university students compared to rank-and-file enlisted men was due to the more intense near work environment associated with advanced study (Ware, 1813). Cohn, in his monograph The Hygiene of the Eye in Schools (1886), attributed the higher prevalence of myopia in urban gymnasia compared to that found in rural schools to the intensity of near work required for their more rigorous curriculums, along with close working distances and poor indoor illumination (Cohn, 1886).

In the 20th century, the classic Inuit studies by Young et al. (1969) documented increasing prevalence of myopia among school-aged children compared to their parents and grandparents, a shift attributed to the increased level of near work that accompanied the children’s newly instituted compulsory schooling (Young et al., 1969; see Chapter 3 for more details). An intense near-work environment in school is still frequently cited as a risk factor for myopia. A classic example is the study of Orthodox rabbinical students conducted in Israel by Zylbermann et al. (1993). The study found that more than 80% of male Orthodox students were myopic compared to less than half that prevalence among their sisters, who were also attending Orthodox schools, and among males or females attending general schools. The near work

demands for Orthodox males, mainly reading and writing, were as high as 16 hours per day, as compared to 2–3 hours per day for the other students (Zylbermann et al., 1993; see Chapter 3 for similar data collected in Israel).

Recent work comparing the prevalence of myopia in different countries and academic test scores from children residing in those countries, using data from the Program in Secondary Assessment (PISA) system created by the Organisation for Economic Co-operation and Development, has confirmed the relationship between education and myopia (Jong et al., 2023). In a statistical model, PISA scores could account for 31% to 64% of the variation in myopia between countries (albeit not addressing whether the relationship was causal in nature). Economic development and urbanization are frequently associated with a higher prevalence of myopia compared to populations living in rural areas within the same country (Ip et al., 2008). The prevalence of myopia typically increases as societies become more affluent; for example, central African countries are predicted to undergo sharp increases in myopia prevalence, from 9.8% currently to 27.9% by 2050 (Holden et al., 2016).

The level of education was associated with increased odds of being myopic, particularly in the presence of risk alleles for refractive error and myopia identified in the CREAM consortium (Verhoeven et al., 2013). A child’s exposure to education is not readily amenable to investigation in randomized trials. Instead, the Mendelian randomization analysis method (Mountjoy et al., 2018) has been used to gauge the potential causal effect of education on myopia: this approach aims, in theory, to reduce the effect of potentially confounding variables, thereby isolating the effects of the environmental variable of interest. An analysis of U.K. Biobank participants showed that each year of additional education was associated with −0.27 D (95% CI: −0.37 to −0.17) less hyperopic/more myopic refractive error; the study had 80% power to detect an effect of time spent in education on refractive error ≥ 0.14 D/year (Mountjoy et al., 2018).

A second research method that has been used to examine the causal effect of education on myopia is regression discontinuity analysis. Plotnikov et al. (2020) took advantage of a ‘natural experiment’ in which a government policy raised the school-leaving age of U.K. children from 15 to 16 years. In those affected by the education reform, refractive error was shifted in the direction of myopia by −0.77 D (95% CI: −1.53 to −0.02 D). In China, the effect of education on myopia has recently been quantified in children using data gathered through nationwide screening programs. One more grade level of education in China was associated with an increase in the prevalence of myopia by a consistent increment compared to children of similar age but with one grade level less of education. This phenomenon was evident at least until the prevalence of myopia began to reach a plateau, in high school (He et al., 2021). A regression discontinuity study of 910,000 children ages 4–14 years living in Shanghai, China, estimated that an additional year of schooling at age 6 years caused a −0.19 D (95% CI: −0.09 to −0.30 D) shift toward myopia, while at age 14 years an additional year of schooling caused a shift of −0.67 D (95% CI: −0.21 to −1.14 D; He et al., 2021).

These independent sets of results from Mendelian randomization and regression discontinuity studies underscore the effect of increasing education on the risk of having a myopic refractive error, but they do not shed light on potential biological mechanisms. Less time outdoors because of more years spent in education may contribute to the link between education and myopia. For instance, when time outdoors was accounted for in a Mendelian randomization analysis of education level, the effect size of −0.27 D for each year of education was reduced by roughly 40% to −0.17 D per year (Clark et al., 2023a). Further research to determine the relative

importance of more near work versus less time outdoors, or other factors, will be important in understanding the link between education and myopia.

Near Work and Myopia Onset vs. Progression

The role of near work in myopia has recently been reviewed by Huang et al. (2015), Gajjar & Ostrin (2022), and Dutheil et al. (2023). However, the role of near work remains contentious (Mutti & Zadnik, 2009), in large part due to the conflicts between cross-sectional and longitudinal studies and the methods used to quantify near work. Near work has traditionally been assessed through surveys of parents estimating the time their children spend in various activities. Having parents estimate children’s time in an activity is simple to incorporate into a survey, but it lacks detail, averages activities into one estimate over perhaps a year’s length of time, asks parents to make an estimate of activity time they have not witnessed in person, and is subject to recall bias. Despite these limitations, some trends have emerged from survey-based research. Working distance and prolonged periods of near work are recurring themes in recent studies that could be evaluated on an individual basis using electronic monitoring technology (Gaijar & Ostrin, 2022; Guo, 2016; Huang et al., 2019). Minimally invasive electronic monitoring technology now permits a child’s working distance and the duration of periods of near work to be evaluated over hours or days (Bhandari et al., 2022; Li et al., 2020; Williams et al., 2019a; see Chapter 4). Working distance and the temporal dynamics of near work, along with more detailed assessments of children’s visual experience in both indoor and outdoor environments, deserve more attention in future longitudinal research.

Notwithstanding the limitations of the methods used to quantify near work, associations between levels of near work and myopia have been reported less consistently in cross-sectional and longitudinal studies than associations between time outdoors and myopia. Cross-sectional studies often evaluate the association between near work and myopia in individuals who are already myopic unless they are undertaken at an early age. Prospective longitudinal studies ask whether increased near work in non-myopic children increases their risk of myopia onset. Here, some of the key findings from these studies are summarized. The CLEERE study (Zadnik et al., 2015) was a multi-ethnic longitudinal study in the United States that monitored refractive development in more than 4,000 children, represented at each baseline age from 6 to 11 years old, with up to 7 years follow-up until the 8th grade. It found that the odds ratio for incident myopia associated with a one-unit increase in ‘diopter-hours’ of near work was estimated to be precisely 1.00 (95% CI: < ± 0.01). In the 3-year longitudinal SCORM study, which followed 994 children residing in Singapore ages 7–9 years at baseline, the relative risk of incident myopia was 1.01 (95% CI: 0.97–1.05) for each additional book read per week (Saw et al., 2006). A one-year longitudinal study of nearly 1,500 Chinese children in grades 1–4 found no association between time spent in near work and increased risk of onset.

In contrast to many negative results for near work and risk of myopia onset in school-aged children, near work exposures prior to school age or early in school may increase risk. In the Generation R study (Enthoven et al., 2020), a birth cohort study that monitored more than 5,000 children in The Netherlands, incident myopia at age 9 was modestly but significantly associated with computer use at age 3 years (OR = 1.005, 95% CI: 1.002–1.010) and computer use at age 6 years (OR = 1.009, 95% CI: 1.002–1.017). In the SAVES study (French et al., 2013), which followed approximately 2,000 Australian children for 5–6 years, younger children in the high tertile for near work had a greater risk of incident myopia (OR = 2.35. 95% CI: 1.30–4.27) while older children in the high tertile for near work had an increased risk that was not

statistically significant (OR = 1.31, 95% CI: 0.83–2.06). In contrast, a recent longitudinal study of initially non-myopic Chinese high school aviation cadets found that spending more than 8 hours in class per day, frequent periods of continuous reading for over one hour, and a near working distance of less than 30 cm—but not reading for more than 4 hours per day—were all significantly associated with a higher incidence rate of myopia over 20 months of follow-up (Yao et al., 2019). This finding is consistent with animal models of myopia that show temporal nonlinearities in the effects of hyperopic defocus, while “total amount of near work” may be a relatively poor measure for exactly these reasons (Wallman & Winnower, 2004). Reports of the association between near work and the progression of existing myopia have been similarly mixed (Dutheil et al., 2023; Gajjar & Ostrin, 2022; Huang et al., 2015). Also discussed below, early exposures to near work may be more significant regarding future myopia progression. However, at later ages, the following studies suggest that once myopic progression begins, near work may have limited effect on progression.

Research by the COMET Group found that near work at baseline had no effect on the proportion of children whose myopia stabilized by age 15, although the amount of myopia at stabilization as a function of near work was not presented (The COMET Group, 2013). A remarkable 23-year follow-up of Finnish participants enrolled at an average age of 11 years found no association between myopia progression and near work assessed in childhood and again at the follow-up visits at either 24 or 35 years of age (Parssinen et al., 2014). A 1-year longitudinal study of nearly 5,000 Chinese children in grades 1–4 found no association between time spent in near work and either axial elongation or change in refractive error (You et al., 2016).

Two studies show an effect on myopia progression from near work. In one study, however, it was an effect of reading distance more than time spent reading, and the effect was small. This two-year longitudinal study of nearly 4,000 Taiwanese children 9–11 years of age examined the association between progression and two near-work behaviors: reading distance and whether near work was uninterrupted for 30 minutes. Refractive errors were compared between groups dichotomized for exhibiting the better visual hygiene for each behavior (reading distance ≥ 30 cm and uninterrupted near work ≤ 30 minutes). There was statistically significantly less myopia in the “protective” group throughout follow-up, although it is unclear if these differences existed at baseline or developed over the course of follow-up. Over the two years of follow-up, the close reading distance group progressed 0.15 D more and the uninterrupted group progressed 0.07 D more than the respective “protective” group children (Huang et al., 2020). These groups were not randomized, and it is unclear how many children were in each group. The other significant association was found in a longitudinal study of 1,279 children ages 5–15 years from India. The odds ratio for progressing in myopia by at least −0.25 D in one year associated with spending more than 42 hours of near work per week compared to spending less than 35 hours was 2.10 (95% CI = 1.24-3.56). Spending more than 7 hours per week using a computer or playing video games was also associated with showing myopia progression (n = 629) vs. having a stable myopic refractive error (n = 650). Time watching television was not significantly related to myopia progression (Saxena et al., 2017).

More detailed assessments of working distance and uninterrupted time in near work seem warranted, as well as longitudinal studies beginning at earlier ages.

A Case for Re-Examining the Theories for the Effects of Near Work on Myopia: Does Defocus Have a Consistent Effect in Driving Eye Growth?

Animal experimental models of myopia produced by lens defocus are often cited as the theoretical framework supporting near work as a major risk factor for myopia incidence. Accommodation in children for near work is typically less than the amount required by the dioptric demand, resulting in under-accommodation, or accommodative lag (Gwiazda et al., 1999; Mutti et al., 2006). The sign of this defocus is hyperopic, with the conjugate point of the near reading material or computer screen in focus behind the retina. The analogous condition in animal experiments is hyperopic defocus imposed by minus lenses, a stimulus that reliably accelerates the elongation of the eyes of young animals across a wide variety of species (Pardue et al., 2008; Schaeffel et al., 1988; Smith & Hung, 1999; Wallman, 1987). The hypothesized connection between animal models and childhood myopia is that prolonged time in near work in the presence of accommodative lag provides the same form of growth signal from hyperopic defocus, thereby increasing the risk of myopia onset.

Accelerated eye growth in response to hyperopic defocus is quite reliable in animals, but the results of interventions in humans that prevent hyperopic defocus during accommodation are disappointing in comparison. If near work and hyperopic defocus were as detrimental to the eye as assumed from animal models, bifocal spectacles would have solved the problem long ago. Clinical trials of segmented multifocal or progressive-addition lenses have failed to produce clinically meaningful results in slowing myopia (Berntsen et al., 2012; Fulk et al., 2012; Gwiazda et al., 2003). The current emphasis in optical treatments is to provide myopic defocus as a ‘stop’ signal in the retinal periphery to inhibit overall elongation while not interfering with visual acuity. These peripheral optical treatments take many forms: contact lenses, overnight orthokeratology, and specialty spectacles (see Chapter 7). The treatment benefit from this approach is greater than from bifocal spectacles, but there are important limitations, as elaborated in Chapter 6. The treatment benefit seems to have a ceiling, limited to 0.50 D to 0.75 D less progression of myopia in treated children compared to controls (Chamberlain et al., 2019; Lam et al., 2020; Walline et al., 2020).

More importantly, the greatest effect is seen early in most myopia control treatments, with diminishing additional treatment benefit in later years. In the BLINK study of multifocal contact lenses, for example, significant inhibition of axial elongation was observed only in the first two of its three years (Mutti et al., 2022). A Cochrane review of myopia treatments shows that this is typical for both optical and pharmaceutical approaches to myopia control (Walline et al., 2011). A child destined to have −6.00 D of myopia may instead become −5.25 D. While this is an improvement, no clinical data exist to support a claim that current myopia control could make the final refractive error of that child −3.00 D. The pattern of inhibition of ocular growth in the BLINK study was also inconsistent with prediction from animal models. Elongation was indeed inhibited at every peripheral point measured out to 30° in the retinal periphery, but the inhibition compared to controls was greatest at the fovea, not in three of the four quadrants of the periphery where myopic defocus was present.

Control of ocular growth by defocus was also inconsistent, with similar amounts of inhibition seen between the vertical and horizontal meridians of the eye despite the substantially greater peripheral myopic defocus vertically (Mutti et al., 2022). Accommodative lag is supposed to be a visual risk factor for myopia onset and progression, but longitudinal results show excess lag in myopes to be more a consequence of myopia rather than a cause (Mutti et al., 2006). Once lag increases after the onset of myopia, the degree of lag is unrelated to the rate of

myopia progression (Berntsen et al., 2011). Hyperopic peripheral defocus is also unrelated to increased risk of onset (Mutti et al., 2011; Sng et al., 2011).

As with accommodative lag, once hyperopic peripheral defocus increases in myopia, studies in the United States, Singapore, the United Kingdom, and China show that the rate of progression is either unrelated or inconsistently related to the amount of relative peripheral hyperopia (Atchison et al., 2015; Mutti et al., 2022; Radhakrishnan et al., 2013; Sng et al., 2011). Animal results clearly point to local control of eye growth through defocus (Gawne et al., 2022; Leng et al., 2010), suggesting that treatments directed at controlling myopic progression through modification of peripheral defocus should be more effective than they currently appear to be. It is difficult to explain why animal experimental results do not translate more effectively when applied to children. Given that clinical interventions based on animal models only show incremental progress against myopia, animal models may not be the strongest foundation for building an argument for near work as a risk factor for myopia onset. At the same time, there is insufficient understanding of how treatments, and even natural near viewing (Labhishetty et al., 2021), affect retinal image quality and defocus under various viewing conditions—whether it is through accommodative effort or pupil size, for example (reviewed in Chapter 7). This gap in knowledge underscores the need to establish more detailed models of retinal image quality during the various visual experiences of childhood when hypothesized associations between environmental exposures and myopia are studied.

Accommodation and Emmetropization in Infancy

Induction of myopia in animal models clearly shows a sensitive period (McBrien & Norton, 1992). The responses to deprivation or defocus are greatest in young animals and diminish with age (Smith et al., 1999; Wallman & Adams, 1987). This age effect could be one reason why interventions that reduce hyperopic defocus in humans are only partially effective in slowing myopia progression: the treated children may be outside of this sensitive period. The effects of accommodation and near work may be overstated in childhood and underappreciated at earlier ages. Emmetropization, the reduction in both the absolute level of hyperopia and the variance in refractive error, is the major characteristic of infant refractive error development (Mutti et al., 2018). Hyperopia in infancy drives eye growth in a negative feedback loop to be self-correcting. The tuning mechanism by which the distribution of refractive error is transformed from a normal distribution at birth into its characteristic leptokurtic shape (far more children than by chance near the average than in a normal [bell-shaped] curve) after the first two years of life is widely considered to be driven by visual feedback. Most researchers working with animal models consider hyperopic defocus to be the operative visual signal.

This hypothesis deserves reconsideration. The accommodative response stimulated by early levels of hyperopia may represent a candidate visual signal for emmetropization. The assumption is that when a negative lens is placed over a neonate eye that the animal is experiencing hyperopic defocus. However, accommodation and the effective refractive state of the animal are rarely assessed. When accommodation was assessed in a key study in macaques, Smith and co-workers made an important statement in their discussion (Smith et al., 1999):

In other words, compensation for imposed hyperopia required a robust accommodative response.

Human infants can produce an adult level of accommodation at 3–4 months of age, which is the age at which the variance in neonatal refractive error begins to narrow (Banks, 1980; Gabriel & Mutti, 2009; Haynes et al., 1965; Mayer et al.; 1995, Mutti, 2007). The effects of defocus and accommodation can be untangled, because they represent equal and opposite sides of dioptric demand. Infants are either experiencing defocus or they are accommodating and experiencing less defocus. The necessary measurements for the analysis are distance refractive error and the accommodative state during a near task. Such an analysis was performed on the 262 infants in the Berkeley Infant Biometry Study. Cycloplegic refractive error was measured at 3 months and again at 9 months, a time period when the majority of emmetropization takes place. After accounting for the effects of underlying hyperopic refractive error, the poorest emmetropization was seen in infants with the greatest amount of hyperopic defocus, both at near and far test distances. More accurate accommodation at 3 months of age translated into more effective emmetropization by 9 months of age (Mutti et al., 2009). Horwood and Riddell found similar results, that non-emmetropizing hyperopic infants under 6 months of age showed high amounts of hyperopic defocus at distance because of poor accommodation, in contrast to the more accurate accommodation of emmetropizing infants (Horwood & Riddell, 2011). Effective accommodation was associated with emmetropization, not hyperopic defocus, in their sample.

Studies that employ spectacle correction of infant hyperopia are difficult to interpret. Spectacles would be predicted to inhibit emmetropization under both theories; correction reduces both hyperopic defocus and the accommodative demand. The key question is the effect of spectacle correction on accommodative response. Unfortunately, this measurement has not been made in these studies. The two major studies of correction of infant hyperopia had conflicting results. Ingram found inhibition of emmetropization after spectacle correction of non-strabismic highly hyperopic infants, while Atkinson found no significant differences between treated and control infants (Atkinson et al., 2000; Ingram et al., 2000).

In sum, accommodation may be an underappreciated potential driver of infant eye growth that deserves greater attention in studies of human emmetropization.

Near Work and Accommodation May Have a Greater Influence on Infant and Early Childhood Refractive Error than Later in Childhood

Accommodation could influence refractive error through the action of the ciliary muscle on ocular shape. An accelerated rate of axial elongation, a distortion in the shape of the eye toward a less oblate shape, and a cessation in the thinning, flattening, and power loss of the crystalline lens all characterize the onset of myopia (Mutti et al., 2007a, 2012). Proportional expansion of the eye axially and peripherally along with compensatory changes in the crystalline lens that have occurred from birth all cease with the onset of myopia. The increase in accommodative lag and in the AC/A ratio (the amount of convergence resulting from one diopter of accommodation) suggest that a possible source of interruption to the necessary changes in the crystalline lens is equatorial restriction from a thickened ciliary muscle (Mutti et al., 2006, 2017). If the ciliary muscle can influence the development of myopia in childhood, perhaps it can be a factor in emmetropization. If the distortion in ocular shape from ciliary muscle tension plays a

negative role in childhood myopia by accelerating axial elongation, it could also be a positive influence in hyperopic infants.

Data supporting the role of accommodation in emmetropization and experimental myopia using animal models are mixed. Chicks, marmosets, and rhesus monkeys all have active accommodation and have been used to investigate the influence of accommodation on emmetropic or myopic eye growth (Chakraborty et al., 2020). Studies have blocked accommodation by lesioning the Edinger-Westphal nuclei, or by performing optic nerve or ciliary nerve section, and have then examined the response to lens-induced myopia (Schaeffel et al., 1990; Raviola & Wiesel, 1990). These studies reported that the response to myopigenic signals remained intact with the lack of accommodation. Additionally, the protective effects of atropine on experimental myopia were first thought to produce paralysis of the ciliary muscle and inhibition of accommodation through atropine’s action as a nonspecific muscarinic antagonist (Chakraborty et al., 2020). However, later studies in chicks found that atropine did not affect the ciliary muscle in the chick, so atropine was likely not preventing myopia by inhibiting accommodation (McBrien et al., 1993).

Another point of dissonance with accommodation contributing to lens-induced myopia is that accommodation changes focus uniformly across the visual field, while regional stimulation of the retina with negative or positive defocus can produce local changes in eye growth (Diether & Schaeffel, 1997). Evidence supporting accommodation in modulating refractive eye growth includes studies showing that brief periods of normal vision during exposure to negative lenses can inhibit experimental myopia (Kee et al. 2007; Schmid & Wildsoet, 1996; Shaikh et al., 1999). Thus, eliminating defocus through accommodation could inhibit myopia development. Further supporting this possibility, accommodative performance before and after inducing experimental myopia in marmosets revealed that increased accommodative lag was present after lens-induced defocus and did not predict the amount of induced myopia (Troilo, 2007). Finally, blocking accommodation in chicks with ciliary nerve section resulted in myopia, suggesting that accommodation may play a role in evaluating the sign of defocus (Diether & Wildsoet, 2005). These conflicting results were summarized by Chakraborty et al. (2020), stating “a complex relationship exists between accommodation and emmetropization, involving multiple neural pathways, feedback loops, and interactions between temporal and spatial patterns of defocus.” Given these various findings, accommodation as a visual signal for emmetropization deserves further study.

If accommodation might reduce hyperopia in a beneficial manner in infancy, how long is this sensitive period? If it extends into early childhood, what would the effect be of intense near work on the eye of a toddler? Gordon-Shaag et al. (2021) studied Israeli schoolchildren and found that the more myopic ultra-Orthodox children learned to read 1–2 years earlier (at an average of 4.3 ± 0.8 years of age) than the less myopic religious or secular students (see Chapter 3). Early computer use at age 3 was associated with later myopia at ages 6 and 9 in the Generation R study (Enthoven et al., 2020). The age at which children learn to read is not a variable typically included in studies of myopia, but it may be one that deserves more attention. The epidemiology of myopia emphasizes this point. The children of East Asia and of Scandinavia have widely different prevalences of myopia. Prevalence in Taiwan may be as high as 80%, while in Scandinavia it may be 16% or lower (Hagen et al., 2018; Lin et al., 2004). “Finnish children first learn the letter sounds at school when they are 7 years old” (Brandslet, 2023). Chinese children in preschool as young as 3 years old may begin to learn to read, and more than 80% of Chinese children attend preschool in a more intensive near work environment,

one with an emphasis on learning to read at an early age (Jiang et al., 2021). In Scandinavia the emphasis in preschool is on play and social interaction, with a considerable amount of time spent outdoors each day regardless of the weather. The data from China suggest that starting school increases the prevalence of myopia relative to children of similar age who begin school one year later, an increase that carries through the elementary grades (Xu et al., 2021).

Summary for Near Work

Addressing near work in childhood is a challenge. Near work is an essential and inescapable part of the reading and learning necessary for productive citizenship. Changing the research emphasis to investigating the effects of near work performed earlier in childhood may be more productive in understanding the role it plays in the onset of myopia. Investigating the effects of early near work, accommodation, and the visual environment of children (including their visual diet in the educational setting) on later myopia development may be a novel and important new area in myopia research.

ELECTRONIC DEVICES

As children increasingly engage with electronic devices at younger ages and their screen time rises, it is imperative to fully understand the association between electronic device usage and ocular health (Liu et al., 2021). Electronic device use has been suggested as one of the environmental risk factors for myopia (Martínez-Albert et al., 2023). Recent studies have shown a significant increase in screen time at earlier first-exposure ages among young children (Byrne et al., 2021; Dumuid, 2020), and even more so during the COVID-19 pandemic (Wong et al., 2021).

For example, the percentages of kids from 8 to 18 years old who own smartphones steadily increased from 2015 to 2021; as of 2021, the 8- to 12-year-olds used screen media about 5.5 hours per day and the 13- to 18-year-olds used it about 8.5 hours per day (see Figure 5-4; Common Sense, 2021). Among households with children younger than age 8, smartphone ownership increased from 41% to 97% and tablet ownership increased from 8% to 75% from 2011 to 2020 (Laricchia, 2022). While the global rise in myopia prevalence predates the advent of smart devices, the recent surge in electronic device usage has been argued to further add to the already high rates of myopia (Dirani et al., 2019; Foreman et al., 2021; Lanca & Saw, 2020). Gaining an understanding of the effects of electronic device use on myopia is crucial for shaping public health policies, educational strategies, clinical practice guidelines, and parenting approaches.

There have been mixed perspectives on the contribution of electronic devices to the increase in myopia (Lanca & Saw, 2020). For example, some vision researchers claim uniquely influential features of using electronic devices, such as different eye-movement patterns induced and light exposure while using them (e.g., Jian, 2022; Miranda et al., 2018), whereas others view electronic device usage as a case of near-vision work, with a byproduct of reduced time outdoors (e.g., Alvarez-Peregrina et al., 2020; Huang et al., 2015). While assessing its impact, it is important to understand the underlying mechanisms associated with electronic device usage.

The following sections will first distinguish among the different types of electronic devices primarily based on the viewing distance, and then discuss the potential mechanisms of any effect on myopia induction and production related to the physical features of the electronic devices and the behavioral consequences of screen use. The chapter then reviews current measures of screen use, related development of recent technologies, and potential influences of the COVID-19 pandemic on screen use and myopia.

Distinguishing Electronic Devices

When discussing electronic devices in the context of myopia, it is critical to distinguish between those with smaller screens (e.g., smartphones and tablets), which are typically viewed close to the observer (like a book) and therefore require active accommodation, versus those with larger screens (e.g., televisions) typically viewed at optically “far” distances. The nature of screen time and consumption in young children has largely changed in the last decade (Milkovich & Madigan, 2020). Recent studies have shown more time spent on near-viewing screens than is spent on far-viewing screens (Cyril Kurupp et al., 2022; Mohan et al., 2022; Wang et al., 2021; Zhang et al., 2021).

It is critical to note that viewing smaller, handheld devices is a type of near work, no different from reading a book in terms of the accommodative state of the eye. Increased near viewing screen time, sustained near work, and decreased time outdoors often go hand-in-hand (Alvarez-Peregrina et al., 2021; Aslan & Sahinoglu-Keskek, 2022; Lanca et al., 2022). An earlier study by Lee et al. (2013) compared the effects on myopia of reading, using a computer, and

watching TV. This survey study of male military conscripts in Taiwan showed that time spent reading was a significant predictor of myopia, whereas time spent using a computer and watching TV were not. More reading time was associated with higher refractive error and computer use was related to axial length, but watching TV was not a significant predictor of either measure. These differences might be due to different accommodation and defocus patterns in these different conditions (Charman, 2011).

In a recent systematic review by Lanca and Saw (2020), studies involving TV screen time were intentionally excluded because they were not performed at near distance. In a 2022 study of 12,241 children in Asian countries from the Sunflower Myopia Asian Eye Epidemiology Consortium, digital screen time use was defined to include the use of TV, smartphone, tablet, and computer. After adjusting for other factors (e.g., parental myopia, outdoor time), screen time, including TV, was not associated with myopia. The authors acknowledge that the inclusion of TV could have changed the results, as TV is viewed at a greater distance than the other digital devices. Thus, when referring to electronic devices or screen time in relation to myopia, it is essential to define the types of devices studied and their typical viewing distances.

Optical Features of Electronic Devices

Electronic device use intrinsically differs from other types of near work in physical characteristics such as the light emitted by the screen—its chromatic content, luminance, and contrast—as well as features such as screen flicker (Loughman & Flitcroft, 2021; Thomson, 1998). These physical characteristics are potential candidates contributing to the effects of screen use on myopia. For example, the light emitted from screens has considerable energy at the blue end of the chromatic spectrum, and this has been of interest in relation to myopia development. Although the sun has much higher unweighted irradiance in blue light than other sources, screens of electronic devices produce high levels of blue light (de Gálvez et al., 2022), more so than other types of near work such as reading given the same level of ambient lighting.

Chang et al. showed that the light-emitting LE-eBooks used by their participants were more short-wavelength-enriched than printed books (see Figure 5-5) depending on the luminant. Other electronic devices show similar spectral peaks, except Kindle, which is not light-emitting (see Table 5-1; Chang et al., 2015). Gringras et al. (2015) measured light emissions of a tablet, e-reader, and smartphone. They also showed that the content displayed on the electronic devices affected the light emission’s intensity, not necessarily its spectrum (see Figure 5-6). With the prolonged periods of screen use and the proximity of screens to our eyes in today’s digital age, users may be exposed to significant amounts of blue light from electronic devices. However, the mechanisms by which blue light affects ocular health and myopia remain unclear (Cougnard-Gregoire et al., 2023; Iqbal et al., 2023). Future studies can focus on the effects of blue light, as well as various spectral profiles, to inform the design of screens and related products (e.g., light-filtering lenses).

TABLE 5-1 Screen Size, Irradiance, and Peak Spectral Wavelength of Print Books and Electronic Reading Devices

A diagonal measurement was used for the screen size dimension. Light readings were taken while devices were at their maximum brightness setting, and all measurements were recorded from the same distance (38-40 cm) from the screen and in the same background conditions. NA, not applicable.

^* Neither the book nor the Kindle eReader emitted light; the irradiance measured was the ambient room light emitted by the ceiling fixtures and reflected by the printed book or the Kindle screen. The iPad, iPad2, iPhone, and iPod Touch are registered trademarks of Apple Inc. The Kindle and Kindle Fire are registered trademarks of Amazon.com, Inc. The Nook Color is a trademark of Barnes & Noble, Inc. Patent Pending.

SOURCE: Chang et al., 2015.

Contrast has been found to significantly influence retinal processing, potentially impacting the onset and progression of myopia. The act of reading a text with dark letters against a bright background primarily stimulates the retinal OFF pathway, resulting in thinner choroids, which is linked to the onset of myopia (Aleman et al., 2018). Conversely, reading text with bright letters on a dark background (i.e., inverted text contrast) produces the opposite effect. These effects have been shown with exposure to text on screens. Swiatczak & Schaeffel (2022) found that inverted text contrast also elicited axial eye shortening for myopic subjects after reading a large screen at a 2-meter distance for 45 minutes. This effect of text contrast was shown to be dependent on text size, only eliciting significant differences in axial length with large text (Swiatczak & Schaeffel, 2022). This effect of contrast polarity has been further shown to have different effects on retinal processing in myopic and emmetropic eyes (Wagner & Strasser, 2023).

These results may guide the designers of screens to consider screen contrast for the purpose of protection against myopia. However, it is important to note that the contrast ratio between the characters and the background also affect legibility and readability. Although white-on-black and black-on-white texts have the same contrast ratio, white-on-black text is more difficult to read than black-on-white text (Proctor & Van Zandt, 2018; Taylor, 1934).

Behaviors Associated with Screen Use

Besides the intrinsic characteristics of screens, screen use is associated with different behaviors that may affect the user’s ocular health, even when performing the same tasks (e.g., reading on different platforms). Bao et al. (2015), working with 120 myopic children aged 6 to 13 years in China, compared their near-vision posture during various tasks: playing a video game on a handheld electronic device, reading on paper, and writing on paper. They found that, although all tasks were performed at a desk, the viewing distance was significantly closer when playing the video game (21.3 cm) than when reading (27.2 cm) and writing (24.9 cm), with head declination being greater when playing the video game (63.5°) than reading (37.1°) and writing (44.5°). Similarly, Read et al. (2023) conducted a controlled experiment to compare the effects of reading platform (smartphone vs. paper) and environment (indoor vs. outdoor) on gaze behaviors for adult myopes and emmetropes. They showed that a closer viewing distance was used for smartphone-based reading than for paper-based reading, and this difference was greater for myopes. The mean 20° peripheral scene relative defocus was also greater for smartphone-based reading, which was likely due to the closer viewing effect and smaller size of the smartphone. One study compared the print-based tasks of writing on paper and reading books with electronic tasks done on an iPad and a cellphone and found no difference between the print tasks and the electronic tasks (Bhandari & Ostrin, 2022). However, their writing task was set up on a desk, while the other tasks were handheld or on the participants’ laps. Thus, further research is needed to understand the differences between using electronic devices and traditional media while performing similar tasks in the context of myopia onset and progression.

Children report that they are drawn indoors by interesting activities, including the use of electronic devices (Larson et al., 2019). Larson et al. (2019) surveyed 543 6th to 8th graders in rural South Carolina and reported that screen time and outdoor time were negatively correlated, and that the amount of screen time was higher than the outdoor time for almost all groups. On the contrary, a survey on 5,844 children ages 9–11 years from 12 countries showed that greater outdoor time spent outside of school hours was associated with higher screen time, with a possible reason being that those who spent more time outdoors were compensated with more screen time (LeBlanc et al., 2015). However, for children at a much earlier age, 2 years, screen time was shown to be negatively associated with outdoor play in 885 children in Japan (Sugiyama et al., 2023). Similarly, a study of 1,772 preschool children in urban and rural China showed that urban children spent more time playing outdoors and less time on screens (Wang et al., 2020). Thus, there seems to be at least a trend in which using electronic devices is associated with less time spent outdoors.

Children can use screens for extended periods of time before they are even able to read, at very early ages. Among 390 children who were ages 2–5 years old in Korea, 31.3% started using smartphones before 2 years of age and 23.4% used smartphones for over an hour on weekends (Chang et al., 2018). Hinkley et al. (2018) found that 575 children aged 2–5 years were reported to have an average of 2.1 hours of screen time per day. Radesky et al. (2020) found that about 15% of children between the ages of 3 and 5 used their mobile devices over 4 hours per day, among the 346 children they studied in Michigan, United States. Xu et al. (2016) study on over 500 children showed that one-year-old children in Sydney, Australia, had an average of 0.64 hours of screen time per day, with 26% having more than one hour per day. The mean screen time increased to 1.37 hours, 2.48 hours, and 2.25 hours for 2-year-olds, 3.5-year-olds, and 5-year-olds, respectively. Moreover, screen time during infancy at age one was found to be predictive of children’s screen time from ages 2 to 5. These behavior patterns, shown in

studies across different countries, are concerning given that rapid eye elongation occurs at young ages.

Measurements of Screen Use

To further the research on the potential impact of screen time on myopia in young children, it is essential to enhance the rigor of measures and reporting. The methodological consideration of how to measure screen time among children has received little attention until recently. Byrne et al. (2021) summarized measurements of screen time in children ages 0 to 6 years old. One finding is a notable increase in articles measuring screen time on mobile devices in addition to TV starting in 2015, which is an indication of increased interest in this topic. Among the 622 articles included in their review, the overwhelming majority (92.4%) of the studies used questionnaires, while the remaining used 24-hour diary logs. In addition, most reported duration of screentime rather than the frequency of screen viewing. Moreover, the majority of the studies did not report psychometric properties, such as the validity and reliability of the measures. It is also notable that device-based usage event monitoring was not used in any of the studies.

In an effort to ensure the psychometric soundness of the measures, Hutton and colleagues (Hutton et al., 2020) developed a composite parent-report measure of screen-based media use (ScreenQ) and provided a psychometric assessment of its validity and reliability. ScreenQ reflects four dimensions cited in current American Academy of Pediatrics recommendations: access to screens, frequency of use, media content, and caregiver-child co-viewing. This measure has been validated among 69 children ages 36 to 63 months old in the United States and among young Portuguese children (from 6 months to 9.9 years old) with strong psychometric properties, including internal consistency, reliability, and concurrent validity (Hutton et al., 2020; Monteiro et al., 2022).

Recent measures for assessing children’s screen usage beyond parent-report surveys are also under development (Milkovich & Madigan, 2020; Radesky et al., 2020). Radesky et al. (2020) used a method called “mobile device sampling” to objectively measure mobile device use among children ages 3 to 5 years old. This method used data gathered by the device operating system (a monitoring app on Android devices and the battery feature on iOS devices) and found that only about 30% of the parents had subjective retrospective reports of children’s device use that were deemed accurate, with 36% deemed to be underestimates and 35% deemed overestimates of their children’s device usage. This device sampling method has a problem, however, in that it cannot accurately measure usage when individuals share devices (Milkovich & Madigan, 2020).

Recent Technologies

In addition to screentime measurement, which is enabled by emerging technologies, measuring screen distance has also become a feature incorporated into mobile devices. For example, the Screen Distance feature (see Figure 5-7 below) is available on iPhones and iPads starting with iOS 17 and iPadOS, as announced by Apple in 2023. This feature, using a TrueDepth camera, detects and alerts users when the mobile device is held closer than 12 inches from the user’s eyes “for an extended period” (Apple, 2024). This feature is based on the assumption that “Viewing a device (or a book) too closely for an extended period of time can increase the risk of myopia for younger users and eye strain for users of all ages.” In addition to

this built-in feature in iOS, mobile applications (apps) are available to remind or force users to hold their mobile devices at appropriate distances (e.g., Samsung Safety Screen available since 2016, the EyesPro App by Parental Control Kroha available since 2020). Potentially, these apps and features could be used for measuring screen time and distance in research to obtain objective measures. Recent studies have also shown the potential of using smartphones as an optical sensing and analysis platform (e.g., as spectrometers), utilizing their advanced on-board sensors (Di Nonno & Ulber, 2021; Fratto et al., 2023). Research has also made it possible to detect when small children are using smart devices, such as through speech and facial features (Li et al., 2013; Qawaqneh et al., 2017), tap and swipe analysis (Li et al., 2018; Vatavu et al., 2015), as well as touch- and sensor-based approaches (Nguyen et al., 2019).

Other recent developments in technologies such as virtual reality (VR) and augmented reality (AR), utilizing head-mounted displays and head-up displays, have been used to convey virtual visual information to operators in applications such as aircraft (Gu et al., 2020), automobiles (Park & Im, 2020), and games (Munsamy et al., 2020). In VR/AR, eye convergence can be manipulated through rotation of the virtual cameras used to present information to the left and right eyes through automatic virtual convergence (State et al., 2001). The convergence distance can be fixed at a predetermined value, such as a few feet away or infinity (Sherstyuk & State, 2010). However, the accommodation distance is determined by the distance of the light rays, which has been measured to be around 3 to 7 meters (Itoh et al., 2021; Kramida, 2015). The mismatch between convergence distance and accommodation distance is called the vergence-accommodation conflict (see Figure 5-8), which is a major source of motion sickness caused by using VR (Itoh et al., 2021).

Recent efforts by both industry and the academy have developed several mitigation solutions to the vergence-accommodation conflict, with examples such as varifocal displays (Stevens et al., 2018), multifocal displays (Rolland et al., 2000), and retinal displays (Topliss et al., 2023). Apple’s VisionPro, a mixed reality headset announced on June 5, 2023, utilizes a retinal display designed to tackle the accommodation-convergence conflict (Topliss et al., 2023). In the VisionPro’s direct retinal projector system developed for the retinal display, light beams are projected to the eyes through scanning mirrors, rather than images being viewed on a screen or surface as in conventional VR/AR systems. However, this direct retinal projector system does not eliminate the vergence-accommodation conflict, although it is believed to “at least partially” (Topliss et al., 2023, p. 4) help eliminate the conflict.

VR/AR is regarded as having the potential to help prevent and manage myopia through accommodation training (Turnbull & Phillips, 2017; Zhao et al., 2018). However, the effects on ocular health of extended use of VR/AR and exposure to the artificial lights in VR/AR displays is under-researched (Jonnakuti & Frankfort, 2023). Despite remarkable advances in these technologies, they have a long way to go before they can ensure proper depth perception, visual comfort, positive user experience, and ocular health (Cebeci et al., 2024; Kramida, 2015).

To utilize the recent technologies for measuring, monitoring, and preventing myopia, a clearer understanding of the mechanisms that cause myopia is needed. Future research is needed to further the potential of electronic-device-based systems to objectively measure myopia and to monitor other environmental risk factors for myopia, as well as design advanced training and prevention techniques.

Impact of COVID-19

“Quarantine myopia” is a term used to describe myopia developed during the COVID-19 home quarantine (Aslan & Sahinoglu-Keskek, 2022). School-age kids (3 to 17 years old) in various countries had increased screen time due to online schooling requirements (Bergmann et

al., 2022; see Table 5-2). Bergmann et al. (2022) found similar patterns during COVID-19 lockdown for younger children from ages 8 to 36 months, who had no online schooling requirements, in 12 countries. They found that longer durations of lockdown were associated with a more significant rise in screen time and that caregivers reported more screen time during the pandemic compared to before.

The latter outcome might be influenced by the age difference in the children studied, as they were older during the pandemic than before. Older children, in general, reported longer screen times compared to their younger counterparts (Bergmann et al., 2022). In China, Sun et al. (2020) found that nearly half of their 6,416 respondents reported increased internet use during COVID-19. Among students in grade schools, more than 50% of students used smartphones for online learning (Wang et al., 2021). Similarly, in India, Mohan et al. (2022) found that 96.7% of their study population, who were 6- to 18-year-olds, used smartphones for online learning during COVID-19. They also showed an association between rapid myopia progression and using smartphones for video games for more than one hour per day. Children had increased use of electronic devices in many countries during the COVID-19 pandemic.

TABLE 5-2 Previous Findings on Lockdown-Related Increases to Children’s Screen Time

SOURCE: Bergmann et al., 2022, under a Creative Commons Attribution-NonCommercial-No

Derivatives 4.0 International License (https://creativecommons.org/licenses/by/4.0/).

Studies conducted and data collected during the pandemic may provide invaluable evidence for the impact of electronic devices on myopia. A systematic review (Cyril Kurupp et al., 2022) on the impact of the COVID-19 pandemic on myopia progression in children, which highlighted reduced outdoor time and increased electronic device use during home confinement, concluded that the heightened usage of electronic devices (e.g., mobile phones and tablets)

among children during the pandemic had a significant impact on the progression of myopia. The review included 10 research papers published between 2019 and 2022, most of them focused on grade school children. However, it is worth noting that most of the studies reviewed by Cyril Kurupp et al. (2022) used questionnaires to study the risk factors of myopia progression, which lack objective measures.

In summary, the causal effect of electronic device usage on myopia is under-explored. The lack of consistency in definitions and measures of both electronic device use and myopia makes it challenging to reach an agreement on whether electronic device usage has a causal effect on myopia. Although recent developments in the technologies of mobile sensors, monitoring and alerting systems, and VR/AR are promising, more research is needed to allow for recommendations regarding the usage and design of electronic devices.

INDOOR VERSUS OUTDOOR VISION

The Nature of the Problem and Its Distal Causes

There are stark differences between the optical environment of the outdoors, where all visual systems evolved, and the indoor setting of the modern home or classroom, which includes a variety of viewing screen technologies. Several lines of evidence indicate that something, or more likely some things, about these differences are causing the worldwide increase in the prevalence and severity of myopia.

Initial studies of this problem were observational and correlational, but nevertheless strongly suggestive. These have been nicely reviewed elsewhere (e.g., Wallman & Winawer, 2004) and summarized in other parts of this report, and so will only be mentioned here. They include the fact that myopia increases with the level of education attained (Goldschmidt, 1968; Sperduto et al., 1983), that cultures in which people still lead outdoor lives have little myopia (Morgan & Rose, 2005), and that the introduction of compulsory education to the children of Native American & Inuit peoples led to dramatic increases in myopia within one generation (Bear, 1991).

More directly relevant to the indoor/outdoor issue, a seminal, large cross-sectional study of Australian school children (Rose et al., 2008) revealed that those who engaged in more near work and spent less time outdoors had, on average, significantly more myopia. And even after stratification for amount of near work and adjustment for other relevant factors, such as ethnicity and parental refractive errors, children who spent more time outdoors were less likely to be myopic. It should be noted, however, that the effect size was rather small, amounting to a difference in spherical equivalent refraction between the two groups of only about 0.3 D. Other observational studies have largely supported the original findings from Australia. For example, a prospective cohort study in Taiwan (Wu et al., 2020) revealed an impressive slowing, and then reversal, in myopia incidence following school-based interventions that promoted time outdoors.

From early observational studies, one of the most relevant analyses for present purposes was performed by Jones et al. (2007), using survey data from the Orinda Longitudinal Study of Myopia to predict which children subsequently became myopic. They performed a logistic regression analysis and identified two key predictive factors: time outdoors and parental myopia history, including a significant interaction between the two such that children at higher genetic risk benefitted more from time outdoors (Figure 5-9). Interestingly, once these two major risk factors were accounted for, the number of hours a child spent reading per week failed to provide significantly additional predictive power. This latter observation, however, could be confounded

by the aforementioned “genetic nurture” if, for example, myopic parents encourage their children to read or otherwise provide an environment that is more supportive of reading (e.g., having lots of books around the house). Similar results were reported in another large longitudinal study of children in the United Kingdom (Shah et al., 2017). The salutary effects of time outdoors were apparent in children as young as age 3, and they remained sizeable (Hazard ratios of ~0.9 per SD of hours outdoors per day) after adjusting for sex, number of myopic parents, and time spent reading.

Subsequent randomized controlled trials (RCTs) have clearly demonstrated a causal benefit of increased time outdoors. But as with previous observational studies, the magnitude of the benefits has been modest. For example, a cluster-randomized (i.e., by school) intervention with 1st-graders in China (He et al., 2015) found that an additional 40 minutes a day of outdoor time produced a decrease in 3-year, cumulative rates of myopia (defined as spherical equivalent refraction of < −0.5 D) from 39.5% in the control group to 30.4% in the intervention group. Again, this is a small decrease (9 percentage point decrease; 95% CI of a 4 to 14% decrease), but it is a convincingly real effect. Likewise, a large cluster RCT in Taiwan (Wu et al., 2018) found

small (but highly significant) beneficial effects of time outdoors after one year: a difference of 0.11 D for children who were non-myopic at baseline, but a larger difference of 0.23 D in children who were already myopic at the beginning of the study. When normalized for the respective pooled standard deviations (Cohen’s d-prime), these amount to effect sizes of 0.18 (generally interpreted as a “small” effect) and 0.6 (a “moderate” effect), respectively. In agreement with this, a recent meta-analysis of five RCTs with a total of 3,014 participants found an overall beneficial effect size of 0.15 D (0.06 to 0.23) and a risk ratio for developing new myopia of 0.76 (0.67 to 0.87).

Taken together, these data overwhelmingly demonstrate that time outdoors matters, particularly when referring to reduction in the risk of myopia onset. Yet the magnitude of the beneficial effects with respect to myopia progression seem rather small. To put them in perspective, these data can be compared to other interventions, including optical interventions using spectacles or contact lenses and medical interventions with atropine eyedrops.

Caveats in Estimating Effect Size

There are several caveats to consider when considering the RCT-based effect-size estimates above. One is that all these studies were done in school-age children—the youngest being 6-year-olds—and mainly focused on the rate of myopia progression, rather than on a delay in onset. Delay in onset has been suggested by some studies (He et al., 2015; Jin et al. 2015; Jones et al., 2007; Wu et al. 2013) as the major beneficial effect of time outdoors. The RCT-based studies thus may be suffering from the same problem pointed out above concerning near work, namely not focusing on early exposure and on prevention. Second, the magnitude of the intended treatment may have been insufficient. Even though children in the treatment group were encouraged to spend more time outdoors, they were still, relatively speaking, spending most of their school day indoors. For example, in the RCT conducted by He et al. (2015), children in the treatment group received an additional 40 minutes of outdoor recess during school, but this is only about 8% of an 8-hour school day. The data shown in Figure 5-9 suggest that 2 hours per day or 14 hours per week provides the maximum effect (Jones et al., 2007).

Finally, over and above the problem of a low dose of outdoors in the treatment group, the actual differences in time outdoors between the treatment and the control groups may not have been as large as one would desire from a scientific perspective. In this regard, the trial reported in Wu et al. (2018) is worth a closer look, as children in both groups wore light meters, giving the investigators an objective measure of light exposure. The investigators reported the number of minutes per week that children spent in four different luminance categories: ≥ 1000 lux, ≥ 3000 lux, ≥ 5000 lux, and ≥ 10,000 lux (Table 4, p. 1246), both at baseline and at the end of the study for both treatment (n = 267) and control (n = 426) groups. The values in Table 5-3 represent a rough integral of light exposure (lux-minutes/week), taken by multiplying the average number of minutes per week that each group spent in each of the four luminance categories and summing the resulting values.

Curiously, both groups rather dramatically increased their total light exposure: the treatment group increased by about 40%, as expected, but the control group’s exposure also increased, by 32%. When Wu et al. (2018) used the raw values for each child—and they had large enough samples for considerable statistical power—they found no statistically significant difference in time spent in any of the four light-exposure categories, either at baseline or during the study, with unadjusted p-values ranging from 0.37 to 0.97 (median, 0.80). It was only when they combined objective data from the light meters with data from self-report diaries that they

found a significant in-study difference, which amounted to about 70 minutes/week more time outdoors for the treatment group.

TABLE 5-3 Integral of Light Exposure.

SOURCE: Wu et al., 2018.

What might have accounted for this puzzling result? As with all behavioral modifications, noncompliance in the treatment group appeared to be an issue (Wu et al., 2018). However, there also appeared to be some “self-treatment” in the control group (Table 5-3). Some of this latter effect may have been due to parallel initiatives launched by the Taiwanese government, “Sport & Health 150” and “Tien-Tien 120,” both of which promoted time outdoors for school children.

In summary, virtually all studies on the effect of time outdoors—both observational and interventional—have found statistically significant, beneficial effects of small to moderate size. The most consistent benefit has been a delay in the onset of myopia, while slowing of progression in pre-existing myopes has been less frequently observed.

How Does Spending Time Outdoors Protect Against Myopia?

In addition to suggesting interventions for myopia prevention, the beneficial effects of time outdoors give us important clues as to the possible “sensed variables” (Figure 5-10) in the visual environment that are most directly relevant to emmetropization. These are variables that may serve as more convenient points of intervention than the modifications in education and behavior entailed by the “increased time outdoors” approach. To take a simple—and currently hypothetical—example: Insofar as differences in the brightness and color spectrum between the indoor and outdoor environments were found to be causal factors, easier-to-implement changes in classroom design (e.g. larger windows to admit more natural light or specific improvements in artificial lighting) might be effective interventions.

One problem is that such a list of the differences in the “visual diet” (Poudel et al., 2023) provided by indoor vs. outdoor experience is long, particularly when possible differences in oculomotor behavior are also considered. Further complicating the problem, there is likely to be redundancy at multiple levels of the emmetropization system, including which specific variables are sensed under different conditions. Moreover, the critical variables may interact with each other in nonlinear ways and may differ, subtly but importantly, in different species. Even so, differences in the visual environment are an important starting point, and the Committee believes that developing sophisticated models of how the environment, whether indoors or outdoors, influences early visual processing will be critical for advancing our understanding of emmetropization (see Chapter 6, Myopia Pathogenesis: From Retinal Image to Scleral Growth). This will require collecting large data sets of the visual environment experienced by observers under different conditions (e.g., Flitcroft, 2012), combined with good models of the eye’s optics (Hastings et al., 2024) and movements (Rucci & Victor, 2018) to determine how the environment is spatio-temporally sampled. Ultimately, the environment’s influence on photoreceptors and the ensuing retinal circuitry relevant to emmetropization will need to be considered. This effort calls for improvements in technology that will allow the longitudinal monitoring of the visual environment at a fine-grained level in children, during the critical period for emmetropization.

One other wrinkle is that the visual diet, regardless of where one is, depends largely on what one is doing. Because most extended periods of near work (e.g., reading) are performed indoors, this introduces correlations among several groups of potential sensed variables. To disambiguate these variables, it would help to find epidemiological circumstances that break these correlations. For example, are there groups of children who spend time indoors but do not engage in much near work? Or, vice versa, are there schools that make use of outdoor classrooms? Below we have focused on the major characteristics of the visual diet that have been investigated experimentally: luminance, chromaticity, dioptric environment, eye movements, and contrast (ON/OFF pathway stimulation).

Effects of Luminance

Because of the remarkable ability of our visual system to adapt to different levels of illumination (Rieke & Rudd, 2009), we are blissfully unaware of even rather large differences: the intensity of outdoor sunlight is roughly 100 times that of a well-lit room (Lanca et al., 2019). This has several consequences for retinal processing that may be relevant to emmetropization and myopigenesis.

There is very strong evidence from studies in animal models that high levels of illumination alone (i.e., absent other features of outdoor vision) are protective against the development of myopia. The ability of bright light to slow eye growth during both form deprivation and lens-induced myopia has been demonstrated in chickens, guinea pigs, rhesus monkeys, mice, and tree shrews (see reviews: Mazade et al., 2024; Muralidharan et al., 2022). It was found in chicks for both form-deprivation myopia (Ashby et al., 2009) and lens-induced myopia (Ashby & Schaeffel, 2010), and in monkeys for form-deprivation myopia (Smith et al., 2012). In the chick form-deprivation myopia model, continuous bright light was more effective than an equivalent amount of intermittent light (Backhouse et al., 2013), and the beneficial effect appeared to be mediated by retinal dopamine (Ashby & Schaeffel, 2010; Muralidharan et al., 2022).

However, it is important to point out that light intensity, per se, cannot be a signal for emmetropization, because it offers no possibility of a differential signal for myopic versus hyperopic defocus that could be used in a negative feedback loop to regulate eye growth. With respect to the benefits of high luminance (outdoors), there are two possibilities: First, high luminance might induce a general signal (e.g., dopamine) that slows eye growth, which would be beneficial in a situation of hyperopic defocus—the starting point for human children (Mazade et al., 2024)—but could be counterproductive under other circumstances. Indeed, Ashby & Schaeffel (2010), using the chick lens-induced myopia model, found that bright light decreased the rate of eye-length compensation for negative lenses and increased it for positive lenses. Alternatively, high luminance might improve the overall operation of emmetropization mechanisms.

The findings of Ashby & Schaeffel (2010) argue strongly for the first possibility. While luminance may not be a signal per se, it can have a huge effect on retinal circuits. Thus, if contrast were an important signal for eye growth, bright light vs. dim light would activate different retinal pathways that could provide different retinal signals for growth.

Because the pupil constricts in response to light, the eye’s optics are improved in bright light: (a) there is better depth of focus, meaning that, other things being equal, an eye with a smaller pupil will experience less blur when viewing objects at different distances from the focal plane and (b) optical aberrations besides defocus and astigmatism are reduced. However, the pupil also constricts as part of the accommodation reflex during near work, so pupil size, per se, may not be a major luminance-related mechanism for the beneficial effects of time outdoors.

Effects of Chromaticity

The sunlight experienced on earth is subject to filtering by clouds and to the effects of Rayleigh scattering in the atmosphere, but under a wide variety of conditions it has roughly equal power at the wavelengths to which the human eye is sensitive (Figure 5-11). In other words, it is “broad spectrum” illumination. In addition, sunlight contains considerable power at wavelengths below 400 nm—so-called ultraviolet light—which are not experienced by the retina due to their absorption by the lens and cornea but affect the production of vitamin D by the skin. By contrast, the spectral composition of indoor lighting varies widely depending on the source (Figure 5-12).

The power of traditional incandescent lighting is heavily concentrated toward the red end of the spectrum, whereas fluorescent lighting tends to be bluer. Increasingly, LED-based lights can have their color spectra tailored from “cooler” (i.e., shorter wavelengths or “bluer”) to

“warmer” (longer/redder). Humans are normally unaware of these wavelength differences, due to extensive spatial-chromatic processing in the retina and beyond. This processing endows our visual systems with the ability to compute the relative reflectance of different surfaces (e.g., the color of paint, independent of the spectral composition of the lighting), largely eliminating the effect of the color spectrum of the light source on our perception of the color of things.

However, just because our visual systems can “discount the illuminant” for color perception, it does not follow that emmetropization has the same capacity or that it does not use chromatic signals that could be affected by the ambient color spectrum. For example, one proposed mechanism by which the retina detects the sign of defocus involves the use of longitudinal chromatic aberration: because shorter wavelengths are refracted by the eye’s optics more than longer wavelengths, hyperopic defocus will result in relatively more retinal blur for longer wavelengths than for shorter; and vice versa for myopic defocus (Figure 5-13).

Thus, a comparison of retinal signals for channels that have different chromatic sensitivities could act as one of the sensed variables for emmetropization. The evidence in animal models is mixed: there is reported evidence that chickens and tree shrews, at least, make use of chromatic cues (Khanal et al. 2023; Muralidharan et al., 2021; Rucker & Wallman, 2009, 2012; Troilo et al., 2019), and also reports that chickens show no effect of chromatic stimulation (Rohrer et al., 1992; Schaeffel & Howland, 1991; Wildsoet et al., 1993). Insofar as spectra from artificial lighting alter the balance in activity between different chromatic channels, they could perturb such a mechanism. This possibility was examined by Gawne & Norton (2020), who performed simulations of the effects on emmetropization of different common sources of artificial lighting using a specific model of longitudinal chromatic aberrations (LCA), the so-called “opponent dual-detector spectral drive model.” The results of their simulations suggested that these artificial sources were chromatically rich enough so as not to perturb emmetropization.

Numerous experiments have been conducted with animals raised under narrow-band illumination, with results varying according to the color of light and the species (see Table 5-4). The consensus from studies in chicks and Guinea pigs is that short-wavelength light promotes hyperopia and is protective against interventions that induce myopia; vice versa for longer-wavelength light (e.g., Foulds et al., 2013; Liu et al., 2011; Seidemann & Schaeffel, 2002; Wang et al., 2018; see Table 5-4 for other references). However, the relationship between wavelength and refractive effect appears to be reversed in rhesus monkeys and tree shrews, with long-wavelength light slowing eye growth (Gawne et al., 2017; Hung et al., 2018; Smith et al., 2015; Ward et al., 2018), and protecting against form deprivation myopia in rhesus monkeys (Hung et

al., 2018; reviewed in Muralidharan et al., 2021). At present, the reasons for the apparent species-related differences are unclear. One important factor that has only recently been identified is the possible differential pre-retinal filtering of light by the cornea and lens. For example, a recent study (Grytz, 2023) found that the tree shrew lens absorbs considerably more violet light than does that of the mouse.

TABLE 5-4 Effect of Raising Different Animals Under Narrow-Band Illumination

SOURCE: Lingham et al., 2020, with permission from BMJ Publishing Group Ltd.

Another route by which the spectral composition of the illuminant might affect emmetropization involves intrinsically photosensitive retinal ganglion cells (ipRGC). In the primate retina, this mechanism begins with a population of giant ganglion cells that express melanopsin, a photopigment whose spectral sensitivity, though complex, is relatively blue-shifted (peak sensitivity at 482 nm), and that appear to roughly (log-)linearly encode retinal irradiance over at least 3–4 log units of intensity (Dacey et al., 2005). These ipRGCs provide

excitatory input to dopaminergic amacrine cells (Roy & Field 2019), the principal source of retinal dopamine, which may be an important molecule in retinal-scleral signaling for emmetropization (see Chapter 6). Sunlight, by virtue of both its broader spectrum and greater intensity, clearly provides sufficient power to stimulate the ipRGC pathway, but certain forms of artificial lighting, such as incandescent lights and “warm” white LEDs, may provide a weaker stimulus. Direct evidence for such a possibility was recently obtained using mouse genetics to target melanopsin or ipRGCs (Chakraborty et al., 2022; Liu et al., 2022). In these studies, investigators eliminated or replaced Opn4 which encodes melanopsin and found altered response to form deprivation or lens defocus myopia, respectively. Furthermore, investigators have found that other atypical opsins, Opn3 and Opn5, which are sensitive to blue or violet wavelengths and are also found in RGCs, might be involved in myopia development. Jiang et al. (2021) found that lens-induced myopia suppression by violet light depended on both the time of day when the treatment was administered and the presence of neuropsin (OPN5). Linne et al. (2023) found that a germline knock-out of Opn3 had more myopic refractions. Overall, this appears to be a promising and important direction for future research.

Effects of Dioptric Environment

The refractive power of the eye’s optics—measured in diopters (D; see definition in Box 2-1)—required to render objects in focus on the retina is inversely related to the distance of the objects targeted. Hence objects at distances less than about 2 meters require positive refraction and, in the emmetropic eye, objects at distances greater than a few meters are essentially “far” (optically speaking, “at infinity”; see Figure 5-14). The term “dioptric environment” is used to denote the distribution of distances of visual surfaces from the observer across the entire visual field.

The outdoor and indoor dioptric environments are radically different (Flitcroft 2012; Gibaldi & Banks, 2019; Marcos, 2024; Sprague et al., 2015; see Figure 5-15). Outdoors, under normal viewing conditions, virtually the entire optical environment is effectively at infinity, whereas indoors there is a much greater range of nearer objects. This dramatic difference could have major implications for the defocus experienced by the retina—particularly the peripheral retina, which for purely geometrical reasons exerts an outsized influence on retinal signals relevant to emmetropization.

With respect to defocus, the animal literature seems to argue that the observer needs to experience nearly constant hyperopic defocus to produce a myopic eye. On the face of it, this would seem unlikely to ever occur in human children, although the indoor environment, with its greater representation of near surfaces, would potentially create greater amounts of hyperopic defocus than the outdoors. This depends, however, on the viewer’s accommodative state. As illustrated by Flitcroft (2012; see panel B in Figure 5-16), a person reading a book in a school or office setting would experience a small amount of hyperopic defocus near the fovea (due to accommodative lag), but the majority of the retinal surface would be experiencing myopic defocus. Based on animal studies of lens-induced myopia, this myopic defocus should slow eye growth and thus be protective against myopia. If the point of regard shifts to a distant object—such as when a student looks at the teacher in the front of the room—then surfaces near the student (her desk and objects thereupon) produce hyperopic defocus, especially in the lower visual field (Flitcroft 2012; see panel A in Figure 5-16).

Flitcroft’s (2012) simulations provide a useful starting point, but they are limited by his use of computer-generated images and by their failure to consider the potential differences in retinal defocus patterns for different observers with differences in eye shape and optical aberrations (for more detail, see Chapter 6). To fully determine the retinal defocus experienced by real children in the indoor environment will require further research to supply (a) comprehensive data sets of the dioptric environments experienced by children during the critical period for emmetropization, (b) appropriate eye models to reveal actual patterns of defocus across the entire retina for differently shaped eyes, and (c) models of fixational eye movements (see below) to characterize the spatiotemporal contrast patterns experienced by the retina by different observers, in different environments, performing different visual tasks.

Effect of Eye Movements

The eyes of foveate animals are in constant motion. At the macroscopic level, primates make several saccades (eye movements between fixation points) every second as they seek to bring high-acuity foveae onto regions of interest in a scene. These relatively large eye movements have the effect of rapidly bringing visual stimuli into the receptive fields of retinal

neurons, producing strong luminance contrast modulation. This is relevant, because experiments in chicks (Rucker et al., 2015) have shown that rapid changes in luminance contrast slow eye growth. Insofar as there are differences in saccade size and frequency in different environments, they could influence emmetropization.

There is precious little relevant data, but at least one study suggests that indoor-vs-outdoor differences in saccade frequency could contribute. Zhang & Vera-Diaz (2016) reported that “Subjects made significantly fewer fixation changes for all indoor tasks (Mean 22 ± 8 fixations), including walking indoors, compared to outdoor tasks (Mean 37 ± 13 fixations; p < 0.01), even for the same task (walking while using an iTouch [a kind of smart watch] indoors vs. outdoors, p = 0.03).” If this result generalizes, it will indicate that more frequent saccades, combined with higher luminance, could be one of the mechanisms by which time spent outdoors has a beneficial effect on delaying myopia onset.

Even when humans believe they are holding their eyes still during periods of fixation, their eyes are in perpetual motion due to so-called fixational eye movements (FEMs), consisting of drift, tremor, and microsaccades. These eye movements have a profound effect on the actual pattern of spatial contrast experienced by the retina and, as such, they are highly relevant to considerations of defocus during emmetropization (Rucci & Victor, 2018). FEMs interact with the spatial frequency distribution of natural images—which have a 1/f² power spectrum (Simoncelli & Olshausen 2001)—to “whiten” the input to the retina. Interestingly, this effect of FEMs could serve to decrease the indoor-outdoor differences in spatial frequency content: The greater high-spatial frequency content of outdoor scenes compared with indoor (Flitcroft et al., 2019) would be low-pass-filtered by FEMs, tending to homogenize the spatial frequency distributions of indoor and outdoor environments. However, experts do not yet know whether there are indoor-outdoor differences in FEMs with respect to their spatial and temporal characteristics. As noted above, such data will be critical for a proper comparison of retinal blur in the two different environments.

Another aspect of myopia research for which FEMs are relevant is the interspecies comparisons implicit in the use of animal models of myopia (Troilo et al., 2019, Wallman & Winawer, 2004). Not much is known about miniature eye movements in other species, but what is known suggests that these movements are quite different in species that lack a fovea (Martinez-Conde & Macknik, 2008). This potentially muddies the interpretation of studies showing, for example, that high spatial frequency environments decrease the development of myopia in chicks (Hess et al. 2006; Tran et al., 2008). Until one knows the spatio-temporal properties of an animal’s eye movements, it is difficult to translate spatial changes in the environment into changes in the optical diet experienced by the retina.

Stimulation of ON vs. OFF Visual Pathways

At the first retinal synapse, between photoreceptors and bipolar cells, visual signal processing is split between a positive-contrast sensing set of channels—the so-called “ON” pathway—and a negative-contrast sensing set of channels—the OFF pathway (Ichinose & Habib, 2022; Schiller, 1992). Because all mammalian photoreceptors hyperpolarize in response to light, the OFF pathway is produced by a sign-conserving, ionotropic glutamate receptor on the dendrites of OFF-bipolar cells, whereas the ON pathway uses a sign-inverting metabotropic glutamate receptor (mGluR6; Figure 5-17). To a first approximation, the two channels can be thought of as full-wave rectifier that converts both light increments and decrements to a positive

signal. This process appears to be fundamental for vision, emerging from models that are built to encode natural scenes efficiently (Gjorgjieva et al., 2014; Karklin & Simoncelli 2011).

However, there are also functional asymmetries between the ON and OFF pathways that are relevant for emmetropization and how it might be influenced by indoor-outdoor differences in the visual diet. The upshot is that a number of the features of the outdoor visual diet, such as higher luminance and higher spatial frequencies, favor the response properties of the ON pathway (Jansen et al. 2019; Luo-Li et al., 2018; Mazade et al. 2019). In addition, certain key players in the retinal-scleral signaling that regulates eye growth, including dopaminergic amacrine cells and ipRGCs, are driven mainly by inputs from the ON pathway (see Retinal Cells and Circuits Regulating Eye Growth in Chapter 6).

With respect to ON vs. OFF pathways, a range of luminance levels may also be important. The outdoor environment contains a broad luminance range (i.e., sunlight to shade), while the indoor environment has a narrow luminance range (i.e., uniform artificial lighting). When testing contrast sensitivity across broad versus narrow luminance ranges using the electroretinogram, ON pathways have a larger response than OFF and this difference increases with luminance range (Poudel et al., 2024). Supporting the benefit of exposure to both dim and bright luminance, exposure to dim ambient light that stimulates mostly rod photoreceptor and

ON pathways may also be protective for myopia in mice (Landis et al., 2021) and children (Landis et al., 2018).

A related issue is the amount of residual retinal motion—i.e., motion not compensated by vestibular reflexes—that is created by movements of the head and body as the observer moves about her environment. This residual motion is a powerful cue to visual stabilization reflexes, such as the optokinetic response, which are driven predominantly by ON pathways (Emran et al., 2007; Sugita et al., 2013; Wang et al. 2023). Insofar as children tend to move around more when they are outdoors (Khawaja et al., 2020), this would present a potentially additional benefit of the outdoor environment for driving ON pathways (Poudel et al., 2023).

These features have led some to propose that a relative under-stimulation of ON pathways during indoor vision may favor the onset and progression of myopia (Poudel et al., 2023). This idea is largely supported by both clinical and basic research findings, which indicate that genetic mutations that selectively degrade ON pathway function are myopigenic, and manipulations that increase ON pathway activation are protective (Crewther & Crewther, 2002; Aleman et al., 2018). For example, patients with congenital stationary night blindness (CSNB) are likely to develop high myopia (Zeitz et al., 2023). And mice with mutations affecting the ON pathway are more susceptible to experimentally induced myopia (Chakraborty et al., 2015; Mazade et al., 2015; Pardue et al., 2008). However, two other studies in which the pathways were manipulated pharmacologically (Crewther & Crewther, 2003; Smith et al., 1991) yielded contrary results, indicating a need for more research (see Chapter 6 for more detail).

Finally, it appears that myopia itself can lead to under-stimulation of retinal ON pathways. A recent study (Poudel et al., 2024) that measured retinal responses using both electroretinography (ERG) and the pupillary light response in human emmetropes and myopes found that as axial eye length increased, the ERG component corresponding to ON-pathway activation decreased. Moreover, luminance increases became less effective at driving pupillary constriction. This result raises the possibility of a pernicious, myopigenic positive feedback loop in which initial myopic changes—whether caused by under-stimulation of ON pathways by the indoor visual diet or other factors—themselves lead to progressively poorer activation of retinal ON circuitry, further diminishing the stop signal to eye growth.

One important caveat to the preceding discussion is that the ON and OFF pathways are each composed of multiple, parallel processing channels, only a subset of which are likely to be relevant to emmetropization and myopigenesis. Because the cleanest distinction between ON and OFF pathways exists at the first synapse, between photoreceptors and bipolar cells, all manipulations performed to date have affected all of the different processing channels that comprise the ON pathways. Given that ON/OFF functional differences may vary for different “pairs” of on-off pathways (Ravi et al., 2018), it will be critical for future studies to home in on the specific inputs that are relevant to myopigenesis. Insofar as particular downstream cell types, such as, for example, dopaminergic amacrine cells and ipRGCs, are implicated in providing signals critical for guiding eye growth, their inputs can be identified using both anatomical and physiological techniques and serve as targets for more specific manipulation.

Sleep Patterns

Human and animal studies have suggested a link between circadian rhythms and myopia (Chakraborty et al., 2018; Hussain et al., 2023). Therefore, it is plausible that a specific aspect of sleep quality also acts as a risk factor for myopia. Recent cross-sectional and longitudinal epidemiology studies have produced mixed results regarding the association between sleep

duration or sleep quality and myopia (Liu et al., 2023). Of the associations identified to date, a shorter sleep duration is the trait that has been most consistently linked to myopia risk, reported in seven of 15 studies addressing this question (Liu et al., 2023). However, shorter sleep duration is also associated with education level, household income, and self-reported race/ethnicity (Whinnery et al., 2014), which means that observational epidemiology studies hold limited scope to address the role of sleep in myopia development, due to the likelihood of bias from residual confounding. Currently, Mendelian randomization cannot be applied to explore the role of sleep duration and myopia, since genetic variants robustly associated with sleep duration in children have yet to be identified; notably, genetic variants associated with sleep duration in adults are not associated with sleep duration in children (Marinelli et al., 2016).

CONCLUSIONS

The recent surge in myopia prevalence has sparked intense interest in discerning its underlying causes, with growing evidence pointing toward environmental rather than solely genetic factors. Shared genetic risk variants across populations with varying prevalence levels strongly suggest the dominant influence of environmental factors on myopia. Notably, inadequate outdoor time among recent generations emerges as a significant environmental variable, with compelling evidence indicating the protective effect of more outdoor time against myopia onset. Studies highlight increased luminance outdoors, likely modulating dopaminergic signaling, as a key factor in this protective effect.

However, further exploration of other environmental disparities between indoor and outdoor settings is warranted. While near work’s role remains less consistently supported, it remains a crucial area for investigation. Disparities in light spectra also emerge as potential influencers of eye growth, although human studies are currently sparse. The “ON/OFF imbalance hypothesis” presents a compelling framework linking visual differences between indoor and outdoor environments to retinal pathways implicated in myopigenesis. Moreover, while the rise of electronic devices coincides with the myopia surge, conclusive evidence regarding their independent role remains elusive, necessitating focused research. However, their association with decreased outdoor time and increased near work underscores the need for nuanced investigation into their impact on myopia risk, especially considering their widespread use among children. Addressing these knowledge gaps is paramount to developing effective strategies for mitigating the myopia epidemic.

Conclusion 5-1: The recent, rapid increase in myopia prevalence over the past few decades, concomitant with heightened industrialization and mandatory primary education, suggests greater weight of environmental influences relative to genetics as causal for this increase. However, genetics contribute strongly to an individual’s susceptibility to environmental risk factors for myopia.

Conclusion 5-2: The environmental variable with the highest level of evidence is the protective effect of time outdoors. The implication is that the prevalence of myopia is increasing, at least in part, because of inadequate time spent outdoors by recent generations of children. Of the features of the outdoor environment that may be beneficial in delaying the onset of myopia, the strongest evidence is for increased luminance, which likely works, at least in part, through dopaminergic signaling. Studies

addressing other salient differences between the indoor and outdoor environments have yet to be tested widely in humans.

Conclusion 5-3: Evidence for the role of near work is less consistent than for time outdoors. The role of near work remains an important gap in knowledge.

Conclusion 5-4: Differences in the spectra of different light sources can clearly influence eye growth in animal models, but studies in humans are currently lacking.

Conclusion 5-5: The “ON/OFF imbalance hypothesis” potentially links many salient visual differences between the outdoor and indoor environments with retinal pathways that have been implicated in myopigenesis, including dopaminergic amacrine cells and intrinsically photosensitive retinal ganglion cells.

Conclusion 5-6: The beginning of the myopia boom preceded the introduction of mobile devices, such as smartphones. Limitations with existing studies make it difficult to determine if use of electronic devices increases the risk of myopia over and above the risk from other near-work activities. This is an important gap in knowledge. Electronic device use does encourage children to spend less time outdoors and more time engaged in near work at close distances for long periods of time and at ever younger ages.

RECOMMENDATIONS

Recommendation 5-1: The Centers for Disease Control should produce evidence-based guidelines, supported by Departments of Education and healthcare providers, promoting more time outdoors (at least one hour per day in school and up to 2 hours total) for children. Consideration should be given to:

• Ensure that outdoor time is safe for the skin and eye by using sunscreen and other protection against short-wavelength exposure.
• Determine the relative importance of more near work versus less time outdoors, or other factors to better understand the link between education and myopia.
• Build comprehensive datasets concerning children’s changing visual diet.
• Include children across the age range of 3 to 16 years.

Recommendation 5-2: The National Institutes of Health and other funding agencies should solicit and fund research to investigate novel questions about the genetic and environmental mechanisms in myopia with special emphasis on the following:

• Studies to identify specific features of the indoor and outdoor visual diet that cause or inhibit myopia development, including potential stimulation through ON and OFF pathways;

• Longitudinal studies of environmental risk factors for myopia that incorporate technologies for capturing data on working distance, temporal properties of near activities, and spectral characteristics of indoor and outdoor activities;
• Experiments in animal models to better understand the mechanisms through which genetic and environmental influences lead to myopia;
• Studies to assess both genetic factors—including polygenic scores—and environmental factors to account for confounding and interactive effects, including better studies of the risk of retinal detachment (etc.) among those with high myopia, by using large cohorts of gene-profiled subjects;
• A single concerted national effort with careful ethical oversight to conduct broad genetic profiling of a large population, combined with detailed health information, that would allow the generation of polygenic scores for myopia (and a number of common diseases) with only small incremental costs;
• The effects of near work, including use of digital devices in preschool children as well as school-aged children, with special attention to the ages at which children are first exposed to these devices;
• The unique risks of developing high myopia, including studies of genetic contributions;
• Better models to understand defocusing across the whole retina as well as fixational eye movement and whether accommodation is a driver of infant eye growth; and
• Better measures of the effects of varying ages when children learn to read.

Recommendation 5-3: Industry partners have an important role in providing:

• Comprehensive quantification of the features of the visual diet of preschool and school-age children;
• Sensors that can be used by researchers to accurately monitor the visual diet of children;
• Research—perhaps in collaboration with academic scientists—on:
  • The visual consequences of the use of their electronic devices, especially in children at risk for the development of myopia;
  • Working distance and the temporal dynamics of near work, along with more detailed assessments of children’s visual experience in both indoor and outdoor environments; and
  • Better comparisons of the differing effects of electronic devices vs. traditional media.

REFERENCES