Association Between Breastfeeding and Likelihood of Myopia in Children

Methods

The study population was taken from all children in grades 1 to 3 of a single junior school in Singapore in 2001, with an 80.2% participation rate. Children with congenital eye disorders were excluded. We conducted a cross-sectional analysis of this cohort in May 2004, when the participants were ages 10 to 12 years. Parents completed questionnaires 1 month before their children’s annual eye examinations. The parents were asked if their child had ever been breastfed and for what duration, and to describe breastfeeding as exclusive, mostly (nonformula supplements), or partly (formula supplements). Parents were not aware of the study hypothesis and breastfeeding questions were mixed with unrelated questions. Research assistants performing the eye examinations were masked to the results of the questionnaire. Parents were aware of their child’s myopia except for 56 children whose myopia was newly diagnosed after completing the questionnaire. We performed cycloplegic autorefraction with an autokeratorefractometer (Canon Inc Ltd, Tochigiken, Japan) and diagnosed myopia as spherical equivalent refraction of at least −0.5 diopters (D). We obtained institutional ethics board approval and parental written consent for all children.

Race/ethnicity was assessed by parental self-report and IQ score was measured by the Raven Standard Progressive Matrices Test.⁴ Proportions and means were compared using χ² tests and t tests, respectively; χ² tests for trend were used for proportions with more than 2 ordered levels. Multiple logistic regression and multiple linear regression models were constructed with myopia and refractive error as the dependent variables, respectively, adjusting for child’s age, sex, race, birth weight, height, books read per week, IQ scores,⁴ mother’s education,⁵ parental myopia, maternal age at delivery, and household income, and accounting for clustering of siblings within families. No interactions were found. Statistical significance was defined as P<.05. Statistical analysis was performed using SAS version 8.02 (SAS Institute Inc, Cary, NC) and Stata version 8.0 (StataCorp LP, College Station, Tex).

Results

Of the 961 children who enrolled in 2001, 797 (82.9%) were reexamined in May 2004. There was no significant difference between participants and nonparticipants with respect to sex, age, or race. Myopia was present in 521 children (65.4%; 95% confidence interval [CI], 62.0-68.7). A total of 418 (52.4%) of 797 children had been breastfed (68 [8.5%] exclusively, 62 [7.8%] mostly, and 288 [36.1%] partly). A total of 298 children (37.4%) were breastfed for 3 months or less; 120 [15.0%] were breastfed longer. The characteristics of children with and without myopia are shown in the Table.

Children who were breastfed had lower myopia prevalence (259 [62.0%] of 418) than children who were not breastfed (262 [69.1%] of 379) (P = .04). A history of being breastfed was associated with decreased odds of myopia (odds ratio [OR], 0.73; 95% CI, 0.54-0.98) and remained decreased after adjustment (OR, 0.58; 95% CI, 0.39-0.84). The mean spherical equivalent refraction of breastfed children (−1.6 D) was less myopic than that of nonbreastfed children (−2.1 D) and remained significant after adjustment (P = .001). Duration of breastfeeding was not associated with myopia (≤3 months vs >3 months; OR, 0.87; 95% CI, 0.51-1.48). The type of breastfeeding (exclusive, mostly, or partly) was not significantly associated with myopia.

Comment

In this cross-sectional study, breastfeeding was independently associated with decreased likelihood of myopia after controlling for known risk factors. Breastfeeding has been linked with neurodevelopment, affecting visual acuity in young children.⁶ Until weaning occurs, breast milk is the main source of many micronutrients including docosahexaenoic acid, which is important for photoreceptor and cortical neuronal development.⁶

Certain limitations need to be considered in interpreting these results. Retrospective collection of breastfeeding data raises a possibility of recall bias, but this is unlikely since parents were unaware of the study hypothesis. As a cross-sectional study, the results indicate association, not causality. Uncontrolled confounding may have been present. Although statistically significant, the ORs have very wide CIs, with the upper end approaching 1. The incidence and progression rates of myopia in Singaporean children, like those of Taiwan and Hong Kong, are among the highest in the world,² about double the rates in the United States,⁷ possibly limiting generalizability.

For these reasons, this study should be considered hypothesis-generating. We recommend prospective studies (cohort or randomized controlled trials) that quantify breastfeeding and evaluate the incidence of myopia in childhood. Confirmation of a protective effect would support the value of breastfeeding.