24 W.C.J. Low, T.Y. Wong and S.-M. Saw

history, near work, outdoor and stature, birth parameters, smoking, and breastfeeding.

Definition of Myopia in Epidemiologic Studies

Refractive error is commonly quantified as spherical equivalent (SE) (sphere + half negative cylinder) in diopters (D) on a continuous scale. Most commonly used and acknowledged definitions of myopia in epidemiologic studies include SE of at least –0.5 D, –0.75 D, and –1.0 D.16 The Refractive Error Study in Children (RESC) used the definition of myopia as SE of at least –0.5 D.17 Other definitions include moderate myopia defined as SE of at least –3.0 D, while high myopia is denoted as SE as least –6.0 D, –8.0 D, and –10.0 D respectively. It should be noted that the cutoff values for myopia are arbitrary and serve to dichotomize the presence of myopia, i.e. myopia present or not present. However, setting an arbitrary cutoff of a physiologic range limits the comparison of studies using dissimilar criteria and disregards the elongation of the axial length (AL). To date, there is no universal accepted definition of myopia.

Risk Factors for Myopia and Ocular Biometry

Family history of myopia

In a population-based cross-sectional study of 2353 Sydney schoolchildren (60% European Caucasian and 15% East Asian) aged 12 years who participated in the Sydney Myopia Study (SMS), children with one and two myopic parents had about two and eight times higher risk respectively (OR = 2.3; 95% confidence interval (CI) = 1.8–2.9 and OR = 7.9; 95% CI = 5–12.4, respectively) of developing myopia (defined as SE at least –0.5 D) compared to those with no myopic parents, after adjusting for age, gender, near work, outdoor activity, and ethnicity (Table 1).18 The level of parental myopia followed a dose-response relationship with children’s myopia onset; increasing severity of parental myopia conferred a greater risk of myopia. The OR for mild myopia (defined as SE from –3 to –0.5 D), moderate myopia (defined as SE at least –6 to –3 D), and high myopia (defined as SE at least –6 D) was 6.4 (95% CI = 1.5–27.8), 10.2 (95% CI = 2.6–40.1),

25 Environmental Risk Factors for Myopia in Children

and 21.8 (95% CI = 5.3–89.4) respectively. However, in SMS, the AL of premyopic eyes did not associate with parental myopia (defined as SE ≤ –0.75D in this analysis).

In a landmark study coordinated by Zadnik and co-workers, (the Orinda Longitudinal Study of Myopia (OLSM) on 716 predominantly Caucasian children aged 6 to 14 years), she demonstrated that the premyopic eyes in children with myopic parents had a longer AL than those without myopic parents, suggesting that the size of the premyopic eyes was already influenced by parental myopia status (Table 1).19 Moreover, she found that children with two myopic parents developed myopia more often (11%) than children with one myopic parent (5%) or children without myopic parents (2%). Myopia was defined as SE at least –0.75 D in this analysis.

In a cross-sectional analysis of 1453 Singapore Chinese schoolchildren aged seven to nine years from the Singapore Cohort Study on the Risk factors for Myopia (SCORM), having one myopic parent increased the AL by 0.14 mm (95% CI = 0.00034–0.25), and two myopic parents increased the AL by 0.32 mm (95% CI = 0.02–0.03) compared with no myopic parent after adjusting for age, gender, books read per week, school, and height (Table 1).20 Similarly, after controlling for the same confounders, having one myopic parent lowered the SE by 0.39 D (95% CI = –0.59– –0.18), and one myopic parents reduced the SE by 0.74 D (95% CI = –0.97– –0.51). The odds ratio of myopia for children with two myopic parents compared with those with one myopic parent was 1.53 (95% CI = 1.16–2.01).

There were other studies that showed the association of family history of myopia with myopia in children, but these studies suffered from methodological limitations such as small sample size, inappropriate sampling strategies, lack of cycloplegic refraction, and lack of control for major confounders.10,21–27 For example, a school-based cross-sectional analysis of 7560 Chinese children aged 5 to 16 years from Hong Kong showed that the number of myopic parents was associated with SE, vitreous chamber depth, and AL in all children (both myopic and non-myopic children) (Table 1).27 However, this Hong Kong study suffered from sampling problems as only selected schools were sampled.

Nevertheless, a previous study demonstrated no significant association of family history with myopia in children.28 In Hong Kong, Fan and coworkers studied 514 Chinese children aged between two and six years but did not find an association of parental myopia status with more myopic refractive error and longer AL (Table 1).28 However, this study is limited by

27 Environmental Risk Factors for Myopia in Children

the school-based design since the schools recruited may not be representative of the general population.

Near work

In a population-based cross-sectional study on schoolchildren recruited in the SMS (n = 2339 and aged 11.1 to 14.4 years), near work parameters were associated with myopia after adjusting for age, sex, ethnicity, school type, parental myopia, and outdoor activity (Table 2).29 Specifically, children who read continuously for more than 30 minutes were 1.5-fold (OR = 1.5; 95% CI = 1.05–2.1) more likely to develop myopia when compared to those who read less than 30 minutes continuously. Likewise, children who performed close reading distance of less than 30 cm were 2.5 times (OR = 2.5; 95% CI = 1.7–4.0) more likely to have myopia than those who performed more than this distance. Similarly, children who spent longer time reading for pleasure and read close at less than 30 cm were more likely to be associated with more myopic SE, after adjusting for age, sex, ethnicity, and school type (p trend = 0.02 and p = 0.0003).

One thousand and five Singaporean children aged seven to nine years were cross-sectionally analyzed in the SCORM; 72.5%, 19.4%, 5.6%, and 2.5% were Chinese, Malays, Indians, and children of other races respectively (Table 2).13 Saw found that children who read more than two books per week were about three times more likely (OR = 3.05; 95% CI = 1.80–5.18) to have higher myopia (defined as SE at least –3.0 D) compared to those who read less than two books per week, after controlling for age, gender, race, night light, parental myopia, and school. Reading more than two hours per day gave a 1.5 times greater odds (OR = 1.50; 95% CI = 0.87–2.55) of having higher myopia compared to those who read less than this amount, but this was not significant. For every book read per week, the AL elongated by 0.04 mm after adjusting for the same covariates. There was a statistically significant interaction effect of parental history of myopia and books read per week on SE (P < 0.001). For example, children with two myopic parents and who read more than two books per week had an age-gender-race adjusted mean SE of –1.33 D, while children with no myopic parents and who read two or fewer books per week had an adjusted mean SE of –0.19 D. A similar effect was found on AL; mean AL of 23.78 mm when the children had two myopic parents and who read more than two books per week vs. mean AL of 23.2 mm in children with no myopic parents and who read fewer than two books per week.