STRENGTHS AND LIMITATIONS OF THIS STUDY
-
A strength of this study was a large dataset, with city-wide-scale investigation.
-
The extensive evaluation of lifestyle risk factors, whether treated as a single factor or combined factor by lifestyle risk score.
-
The effects on myopia were explored comprehensively through analysing the interaction of eye-healthy lifestyle and genetic factors, and the influence of paternal myopia and maternal myopia on myopia was distinguished, separately.
-
Information bias was inevitable due to self-reporting lifestyle risk factors.
-
Non-cycloplegia refraction was tested during ocular examinations, which could result in an overestimation of myopia prevalence.
Introduction
Myopia is a common eye disorder that develops during childhood and early adulthood. It occurs when excessive elongation of the eye causes images of distant objects to focus in front of the retina and contributes to unclear distance vision.1 Myopia has emerged as a significant public health concern, and its prevalence is still rapidly increasing. It is estimated that myopia is likely to affect approximately 50% of the worldwide population by 2050.2 High myopia is particularly concerning because it is linked to significantly higher rates of visual impairment and blindness via retinal complications such as myopic macular degeneration.3–5 In 2015, the global potential productivity loss was estimated at US$244 billion due to uncorrected myopia and US$6 billion due to myopic macular degeneration.6 Meanwhile, the cost of correcting myopia and treating its complications could impose a significant financial burden on individuals and society. Therefore, knowledge of the factors affecting myopia is crucial.
Genetic and environmental factors play a central role in the onset and progression of myopia. The heritability of myopia has been investigated through familial and genome-wide association studies. Twin studies show the highest estimates of heritability of refractive error varying from 50% to 96%, followed by that of familial aggregation studies.7–10 Children with two parents with myopia are five to seven times more likely to be myopic than those with one or no parent with myopia.9 10 However, the independent impacts of the environmental factors cannot be sufficiently controlled in this research. When taking into account environmental variance, a majority of heritability remains unsettled. The growing body of evidence indicates that environmental factors are associated with myopia. Established evidence supports that outdoor activity protects against myopia development in both animal models and human studies.11–13 Additional factors affecting myopia include near work time, reading posture, sleep time, green space and eating habits.14–19 The role of extrinsic factors in the progression of myopia cannot be understated.
The rapid increase in the prevalence of myopia is likely explained by a complex interaction between lifestyle and heritability. Heritability may be acquired by assessing parental myopia status. However, research on the interaction between parental myopia and lifestyle in children has been limited. It is unclear whether individuals might be genetically liable to develop myopia if exposed to certain lifestyles of myopia. Thus, research exploring the differential effect of lifestyle on heritability remains a worthy goal. This study aimed to examine whether eye-healthy lifestyle can influence parental myopia’s effect on children aged 6–18 years developing myopia and to provide new information regarding this critical public health issue for prevention.
Methods
Study population
This cross-sectional study used the Tianjin Child and Adolescent Research of Eye (TCARE) data. The TCARE is a city-wide, school-based and large-scale ocular examination and survey conducted annually. This study was performed on children and adolescents in primary and secondary schools in Tianjin, China, from August to October in 2022. A stratified, multistage, cluster sampling design is used to select participants in the survey. First, the probability proportional to size sampling method was used to randomly select three districts in each rural, urban and suburban area, totalling to nine districts. Second, 20 primary schools, 10 junior high schools and 6 senior high schools were randomly selected from each district, respectively. Third, 2nd, 4th, 6th, 8th and 11th grade were randomly selected from each school. A total of 49 035 participants between 6 and 18 years of age completed the questionnaire; of them, 32 229 participants underwent ophthalmological examinations. 16 806 students failed to receive an ophthalmological examination because they completed distance education at home during the COVID-19 pandemic. Among 32 229 participants, 94 children with missing data for parental myopia, 161 children with missing data or abnormal data for homework time, and 135 children with missing data or abnormal data for outdoor exercise time were excluded. Finally, 31 839 participants were analysed. A flow diagram of participants can be found in figure 1.
Flow diagram of participants in this study.
Ocular examinations
Visual acuity was measured as uncorrected visual acuity (UCVA) using a standard logarithmic visual acuity E chart at a distance of 5 m. Corrected and uncorrected visual acuity was measured separately for those with spectacles. Refraction was measured without cycloplegia with the Tianle RM-9600 autorefractor (Shanghai, China). Non-cycloplegic refraction was performed three times in each eye on each participant, with the average value being used. Each individual was re-examined if there were more than 0.50 dioptres (D) differences between any two of the three outcomes. Ocular examinations were performed by certified ophthalmic practitioners who had received standardised training. All the data were uploaded to an internet access database instantaneously.
Spherical equivalent refraction (SER) was calculated as the spherical dioptres plus half of the cylindrical dioptres (sphere+1.5 cylinder). Myopia was defined as an SER of ≤−0.50 D and a UCVA <5.0 in either eye according to the appropriate technical guidelines for preventing and controlling myopia in children and adolescents by the National Health Commission of the People’s Republic of China.20
Measurement of covariates
Demographic characteristics, behaviour and lifestyle and parental myopia status were collected through a Child and Adolescent Behavior Questionnaire filled in by the parents. Demographic characteristics included age, sex, region, weight, height, etc. Behaviour and lifestyle included constant near work, reading distance (<33 cm or ≥33 cm), poor reading posture, sleep deprivation, homework time (<2 hours/day or ≥2 hours/day), outdoor exercise time (<2 hours/day or ≥2 hours/day), frequency of green space per month (<2 times or ≥2 times), intake of salt (<4 g/day or ≥4 g/day) and intake of carbonated drinks or milk tea (≤1 time per month or >1 time per month). For constant near work, the question ‘Will you take a long look or close your eyes to rest after 30 min of near work?’ was asked. Poor reading posture was defined as reading while walking or in a moving car. Sleep deprivation was assessed based on specific criteria for primary (<10 hours), middle (<9 hours) and high school students (<8 hours).21 For parental myopia, paternal myopia and maternal myopia were registered by a questionnaire separately.
Lifestyle risk score of eye health
Lifestyle risk score (LRS) of eye health was calculated based on each lifestyle factor’s beta-coefficient (β) in the multivariate regression model, adjusting age, sex, region and parental myopia. Backward logistic regression analysis was performed until the final model only included significant lifestyle risk factors (p<0.05). Each lifestyle risk factor was assigned 0 or 1. Each binary lifestyle variable was multiplied by the β, summed, divided by the sum of the β and multiplied by the number of risk factors.
Classification of parental myopia
Parental myopia was categorised into four groups: ‘0’ (no parents with myopia), ‘1’ (paternal myopia), ‘2’ (maternal myopia) and ‘3’ (both parents with myopia).
Statistical analyses
Normally distributed continuous variables were presented as mean±SD, while skew continuous variables were described by medians and quartiles (P25, P75), and categorical variables were shown as frequencies and percentages. The t-test or χ2 test was used to examine the differences in essential characteristics between individuals with and without myopia. Multivariate logistic regression was used to analyse the associations of lifestyle risk factors and parental myopia with myopia, controlling for age, sex and region. The interactions between LRS and parental myopia were also analysed. Parental myopia (‘0’, ‘1’, ‘2’ and ‘3’) and lifestyle risk tertiles (low, medium and high) were converted into 12 dummy variables. These were then used in the multivariate logistic regression models to obtain β, ORs and covariance. Three indicators for interaction evaluation were determined by β and covariance: relative excess risk due to interaction (RERI), attributable proportion due to interaction (AP) and synergy index (S). The bootstrap resampling method was used for calculating the corresponding 95% CIs. The sensitivity and specificity of different predicted models were calculated, and the overall effectiveness was analysed by receiver operating characteristic (ROC) curves. Multiple linear regression was used to evaluate the associations between lifestyle risk factors and parental myopia with SER. The SER of the right eye was used to analyse, as Pearson’s correlation coefficient for the right and left eye spherical equivalent was strong (r=0.861, p<0.01). A two-sided p value of <0.05 was considered statistically significant. Statistical analyses were applied using SPSS software V.24.0 and R software V.4.2.3. Figures were created using R software V.4.2.3 and Origin software V.2021.
Patient and public involvement
Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Results
This study included 31 839 participants aged 6–18 years old, and the overall prevalence of myopia was 55.46%. The mean age was 10.98±2.77 years. Among the participants, 16 643 (52.27%) were boys and 15 196 (47.73%) were girls. The characteristics of participants with and without myopia were shown in table 1.
Characteristics of participants with and without myopia
A backward regression model indicated that reading distance, poor reading posture, sleep deprivation, homework time, outdoor exercise time, frequency of green space, salt intake, and carbonated drinks or milk tea were significantly associated with myopia (p<0.05). For parental myopia, paternal myopia (OR=1.451, 95% CI: 1.346 to 1.564), maternal myopia (OR=1.525, 95% CI: 1.426 to 1.631) and both parents with myopia (OR=2.101, 95% CI: 1.973 to 2.237) were associated with a higher risk of myopia, compared with no parents with myopia (table 2).
Logistic regression analysis of the correlation of parental myopia and lifestyle risk factors with myopia
For interactions between LRS and parental myopia, the 95% CI of RERI and AP did not contain 0 (RERI=0.836, 95% CI: 0.518 to 1.295; AP=0.380, 95% CI: 0.288 to 0.464), and the 95% CI of S did not contain 1 (S=3.288, 95% CI: 2.597 to 4.556), indicating that the additive interaction was significant for myopia. Figure 2 suggests that the risk of myopia among individuals with two parents with myopia and the highest tertiles for lifestyle risk was increased (ORcombined=2.979, 95% CI: 2.681 to 3.311), and was higher than the multiplication of the risks among participants who had only one of these factors (ORcombined for both parents with myopia=1.989, 95% CI: 1.806 to 2.190; ORcombined for high LRS=1.259, 95% CI: 1.142 to 1.388) (figure 1 and online supplemental table 1). The predictive value (calculated as the area under the ROC curve) for LRS & parental myopia was 0.714 (95% CI: 0.709 to 0.720), which was higher than LRS (0.693, 95% CI: 0.687 to 0.699) and parental myopia (0.710, 95% CI: 0.704 to 0.716) separately (figure 3). After the post-hoc tests, the results of the pairwise comparison showed that all p values were less than 0.001.
Supplemental material
OR for myopia per parental myopia and lifestyle risk of myopia tertiles. The depth of the colour represents the size effect. *P<0.05. R, reference.
ROC plots. Reference model included age, sex and region; LRS, parental myopia and LRS and parental myopia model adjusted for age, sex and region. AUC, area under the curve; LRS, lifestyle risk score; ROC, receiver operating characteristic.
The multiple linear regression analysis results showed the associations of lifestyle risk factors and parental myopia with SER, after adjusting for age, sex and region. Constant near work and poor reading posture failed to reach statistical significance (β=0.042, 95% CI: −0.004 to 0.088; β=−0.050, 95% CI: −0.108 to 0.007). Paternal myopia (β=−0.433, 95% CI: −0.495 to −0.371), maternal myopia (β=−0.429, 95% CI: −0.485 to −0.374) and both parents with myopia (β=−0.763, 95% CI: −0.814 to −0.712) were associated with worse SER, compared with no parents with myopia (table 3).
Multiple linear regression analysis of the correlation of parental myopia and lifestyle risk factors with SER
Discussion
In this cross-sectional study of 31 839 Chinese children and adolescents aged 6–18 years, we found a myopia prevalence of 55.46%. There were additive interactions between parental myopia and LRS of eye health, indicating that individuals with a high LRS in combination with worse parental myopia status had a greater risk of myopia than those with only one of these factors. The effect of interaction on myopia increased step by step within higher levels of parental myopia in combination with higher levels of LRS. The predictive value for LRS & parental myopia was 0.714, significantly higher than LRS and parental myopia separately.
This study investigated the effect of parental myopia and eye-healthy lifestyle on myopia outcome as single exposures. Our study suggested that children with parents with myopia increased myopia risk consistent with established research. A study on children aged 12–13 years in Northern Ireland concluded that children with one or both parents with myopia were 2.91 and 7.79 times more likely to develop myopia than those with no parent with myopia, respectively.22 Our study not only investigated the effects of parental myopia, but also analysed the effects of paternal and maternal myopia on children’s myopia. Our survey showed that the probability of having myopia for children with maternal myopia increased by 7.4% compared with those with paternal myopia. The study of Bian et al also showed that maternal myopia increased the risk of myopia than paternal myopia.23 This also reminds us to pay attention to children whose parents have myopia and underscore maternal myopia. Generally speaking, mothers spend more time with their children than fathers do. Mothers with myopia are more likely to have higher educational levels and pay more attention to their children’s study, so that their children spend more time on studying. Thus, genetic factors and the lifestyle of mothers with myopia jointly increase the risk of myopia in children.
With respect to lifestyle factors, we found significant associations for reading distance, poor reading posture, sleep deprivation, homework time, outdoor exposure, frequency of green space, and intake of salt and carbonated drinks or milk tea with myopia. These factors remained significant after adjustment for age, sex, region and parental myopia, suggesting that they are independent risk factors. This is in line with previous findings.24–27 Yang et al examined the effect of varied reading distances on the accommodation of children and observed accommodative lag for near distance. With decreased reading distances, accommodative lag increases, which may induce hyperopic retinal defocus and may aggravate myopia progression.28 The mechanism for myopia due to excessive near work is similar.29 Bright outdoor light might prevent myopia via stimulating the release of dopamine from the retina, because dopamine has been proven to inhibit axial elongation.30 In an animal model using chicks, sunlight may have a protective effect against the development of myopia.31 This may be the possible reason for outdoor exposure and green space as protective factors for myopia. Myopia is caused by sleep deprivation because the circadian rhythm of children is disrupted, which may affect the growth of the eyeball.32 Previous study supports that long-term high-glucose intake leads to chronic elevation of plasma glucose, insulin and serum triacylglycerol levels and then promotes the transmission of various metabolic pathways, which can affect myopia progression.33 In China, students with high academic performance are selected for higher-level classes under an examination-driven educational system.34 Chinese students are pressurised to study; therefore, they tend to spend large amounts of time on homework leading to reduced outdoor and green space exposure, or even less sleep time, causing the risk of myopia to be further increased. Our study highlights the importance of lifestyle adjustment to reduce the risk of developing myopia. Given that eye habits are multifactorial, our study used an LRS of eye health to assess the overall effect of behavioural habits on myopia, which is more comprehensive than previous studies, evaluating the effect of a single factor on myopia and the interaction of factors may be overlooked.
This study concluded that there was an additive interaction between parental myopia and eye-healthy lifestyle. Some scholars have proposed that exploring biological interaction using the additive model is more scientific.35 In our analysis results, familial risk is influenced by lifestyle exposure, and vice versa. Children with high lifestyle risk had a higher risk of myopia than those with low lifestyle risk if they were exposed to the same familial risk. Of all the combinations, children with two parents with myopia in combination with high lifestyle risk had the greatest risk of myopia. This would imply that we should pay more attention to students whose parents have myopia to actively correct poor lifestyles to reduce the possibility of developing myopia.
As molecular genetics and related detection technologies advance, a large number of genes related to myopia have been identified. Currently, genetic testing for children is not feasible in a clinical environment, nor at general population screening. Hence, the assessment of genetic risk score is unlikely to become a routine procedure. The study of Enthoven et al revealed that the predictive value for parental myopia (0.67) was as good as the genetic risk score (0.67).36 Our study showed that the predictive value for LRS & parental myopia was up to 0.714. It is significantly easier to detect children at risk of myopia before its onset by ascertainment of parental myopia and LRS of eye health. Given that changing habits is relatively difficult, identifying high-risk individuals with myopia may specifically help children at risk to adhere to lifestyle advice.
The strengths of this study included a large dataset, city-wide-scale investigation and the extensive evaluation of lifestyle risk factors, whether treated as a single factor or combined factor by LRS. The effects on myopia were explored comprehensively through analysing the interaction of genetic factors and eye-healthy lifestyle. For genetic factors, the influence of paternal myopia and maternal myopia on myopia was distinguished separately. This study also had several limitations. First, the correlation was found in the cross-sectional study, but not causation. Second, UCVA and non-cycloplegia refraction were tested during ocular examinations due to the large number and considering the feasibility of the survey. This could result in an overestimation of myopia prevalence.37 Third, the exclusion of eligible children from final analyses (children without refractive error data or complete lifestyle factors data) may have inevitably introduced selection bias. Fourth, recall bias was inevitable due to self-reporting lifestyle risk factors. Besides, in self-report questionnaires, study participants have a tendency to answer questions in such a way as to present themselves in socially acceptable terms, or in an attempt to gain the approval of others, which could introduce social desirability bias. Future studies incorporating more accurate evaluations of lifestyle risk factors and cycloplegic refraction should be considered in a small sample size. Finally, the study was conducted in Tianjin, China, which may limit the generalisability of the findings to other populations with different lifestyles and genetic backgrounds.
Conclusions
In conclusion, high-risk lifestyles of myopia and parental myopia were significantly associated with a higher risk of myopia in school-age children, and these both had additive interactions. Our findings highlight the need for teachers and parents to promote eye-healthy lifestyle interventions, and children with parental myopia should be targeted during myopia prevention efforts.
Data availability statement
Data are available upon reasonable request.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants. The study protocol was approved by the Medical Ethics Committee of Tianjin Medical University Eye Hospital, China (no. 2020KY-39) and the Ethical Committee of Tianjin Medical University General Hospital (no. IRB2022-YX-087-01) and conducted according to the Declaration of Helsinki. Informed consent to participate was taken from parents/legal guardians of minor participants. The form of informed consent is oral without signature. The Medical Ethics Committee of Tianjin Medical University Eye Hospital exempted informed consent signature. In addition, a professional institution is responsible for data management and data security.
Acknowledgments
We appreciate all of participants who enrolled in this study.
This post was originally published on https://bmjopen.bmj.com