In September 2020, we conducted a cross-sectional study based on a multistage stratified cluster random sampling in Huantai City, Shandong Province, China. Total of 4,494 children from 9 schools (2 kindergartens, 4 primary schools, 2 middle schools, and 1 high school) participated in the survey. Children with eye diseases, such as nystagmus, fundus diseases, and any history of eye surgery, were excluded.

All children underwent examinations at school, including an interview with a standardized questionnaire similar to the one used in the RESC (Refractive Error Study in Children) studies. The questionnaire included basic demographic data such as age, gender, parental refractive status, and lifestyle parameters such as indoor near work time and outdoor activity time.

Firstly, two ophthalmologists performed slit-lamp examinations and fundus examinations for all children before the study, to exclude children with eye diseases that affected this study. All other examinations were performed by professional optometrists, including habitual visual acuity (HVA) at a standard testing distance of 3 meters (#600722, Good-Lite Co., Elgin, IL, USA), non-contact tonometers (Topcon CT80; Topcon Corp., Tokyo, Japan), non-cycloplegic and cycloplegic autorefraction (Nidek ARK-1, CO., LTD, Japan) and Stereoacuity test (Titmus test, described below). Axial length (AL) was acquired using laser interferometry (IOL-Master 500, Carl Zeiss Meditec AG, Jena, Germany), measured 3 times, and the average of readings was calculated. The cycloplegic eye drops used in this study were 1% cyclopentolate hydrochloride Ophthalmic Solution (Alcon, Fort Worth, TX, USA), instilled once every 5 min for a total of 3 times. Then the subjects closed their eyes and rested for 30 min before observing pupil light response and diameter. If the pupillary light reflex disappeared or the pupil diameter was > 6 mm, an autorefractor was performed. If the pupil diameter is < 6 mm, another eye drop was put in. The cycloplegic refraction was acquired 10 min later.

Stereoacuity was assessed by using the Titmus test (Stereo Optical Co., Inc., Chicago, IL, USA). Titmus test ranged from 800 to 40 sec of arc. Wearing polarized glasses, the subjects viewed the fly at a distance of 40 cm and were asked to “pinch” the tip of a wing between the thumb and forefinger. If successful, they were asked to point to the circle that appeared ahead of the plane and seemed to come closer to them. If the subject made one mistake, the previous target was reassessed. If the subject made an accurate judgment on the previous one, we recorded the level of stereopsis. Other detailed examinations have been reported in previous studies (6).

Spherical equivalent (SE) was defined as the sum of the spherical refractive error plus half of the cylindrical (CYL) refractive error (measured as minus values). SE and CYL anisometropia was defined as the difference in cycloplegic inter-eye SE or cylinder degree≥1.00 D, respectively. The differences in inter-eye SE and CYL were presented in the form of absolute values. According to the International Myopia Institute (IMI) (7), myopia was defined as SE ≤ −0.50 D after cycloplegia. Premyopia was defined as −0.50 D <SE ≤ 0.75 D after cycloplegia, and hyperopia was defined as SE > 0.75 D after cycloplegia. Myopic anisometropia was defined as an unequal amount of myopia in both eyes, and hyperopic anisometropia was defined as an unequal amount of hyperopia in both eyes. The worse eye refers to refractive error of the eye with the highest absolute refractive error. If both eyes had the same SE, the right eye was defined as the worse eye. According to the average of cycloplegic SE [(SE of the right eye + SE of the left eye)/2], the refractive status was divided into five groups, including moderate to high hyperopia (SE>2.00 D), low hyperopia (0.75 D<SE ≤ 2.00 D), premyopia, low myopia (−3.00 D <SE ≤ −0.50 D), and moderate to high myopia (SE ≤ −3.00 D).

All data were analyzed using the SPSS 25.0 software (Inc Chicago, IL). Measurement data were given as mean ± standard deviation (M ± SD) and median. Enumeration data were expressed as cases/percentage [N (%)] and 95% confidence interval (CI). Independent-Samples t-test was used to compare the differences between groups. Categorical data were analyzed using the Chi-square test for variance and trend test (P_trend). Continuous data were tested using polynomial linear correlation in one-way ANOVA for differences between multivariate groups and trend test (P_trend). The influencing factor analysis uses Variance Inflation Factor (VIF) diagnostics to diagnose the multicollinearity of independent variables. The Generalized Linear Model (GLM) was used for multivariate analysis of all ocular parameters and lifestyle parameters with statistically significant differences (P < 0.050) in univariate analysis and VIF <4. The effect size of the covariates was based on GLM coefficients (β) with a 95% CI. Under a logistic model, odds ratio (OR) and 95% CI were computed to assess associated factors for the prevalence of anisometropia. All P-values were 2-sided and were considered statistically significant when the values were < 0.05.

We found that anisometropia was more common in hyperopic and myopic children, and more myopic and hyperopic refractive errors in one eye were associated with the occurrence of anisometropia, which has also been validated in other studies (10, 11). The greater difference in the speed of emmetropia and myopia, the greater the possibility of anisometropia, which may be due to the failure of the regulation of the inter-eye homeostatic mechanism (12). It was suggested that preventing myopia and early treating anisometropic amblyopia could be an effective way to reduce the prevalence of anisometropia.

The difference in inter-eye AL was associated with the occurrence of anisometropia in our study, which confirmed that the unbalanced development of two eyes was the main reason for anisometropia in the previous study (13). A cross-sectional study (13) found that differences in the growth rate of the AL were the main factors in the development of anisometropia, which validated our conclusion. Astigmatism in children is mainly from corneal astigmatism (CA). The difference in inter-eye CA may be attributed to increased AL and changes in posterior scleral structure affecting the cornea at the limbus (14). The difference in inter-eye CA would lead to the occurrence of anisometropia and the further development of the difference in inter-eye SE. The results showed that the larger difference in inter-eye CA was associated with anisometropia in our study. In other words, if children with differences in inter-eye AL and CA were found, special attention should be paid to the possibility of anisometropia in clinical diagnosis.

In our study, after adjusting the refractive error and difference in inter-eye AL, the difference in inter-eye HVA was strongly associated with anisometropia. HVA was more conducive to visual performance in everyday conditions (either with or without optical correction) (15). The difference in inter-eye HVA≥0.2 logMAR indicated a higher risk of anisometropia. Previous studies also have shown that the severity of anisometropia was positively correlated with visual impairment (16, 17). Retinal image blur and unequal images caused by anisometropia were the main reasons affecting stereoscopic vision (17). Therefore, when these symptoms are found, it is recommended that ophthalmologists and optometrists take measures to intervene in advances, such as eliminating the form deprivation and abnormal interaction between the eyes (eliminating the inhibition of non-amblyopic eyes on amblyopic eyes).

Goldschmitt (18) speculated that environmental factors might affect the symmetry of binocular refraction. Some etiological studies (9, 19) also found that long time spent in near work was a risk factor for anisometropia. In our study, after adjusting for factors such as age and AL, the larger anisometropia was associated with indoor near work time per day on weekends and outdoor activity time per day on weekdays, which was consistent with our previous findings (4).

When children engaged in near work, such as daily reading and writing, the eyes with more myopic refraction were better to work at a near distance [20]. The difference between inter-eye accommodation demands may provide stimulation for the asymmetric growth of the eyes, which could accelerate the process of myopia in the worse eye, resulting in a larger difference in inter-eye SE and the occurrence of anisometropia. For example, when the working distance was close and the head was tilted, it was assumed that the eyes with lower binocular accommodation requirements had a consistent adaptive response. It might lead to hyperopic defocus in eyes with lower accommodation demand (9). The state of the object image focused on the retina was different. The differences can result in different image quality, which could provide stimulation for the occurrence of anisometropia. Thus, indoor near work time and outdoor activity time may be associated with the occurrence of anisometropia. Children's lifestyles are difficult to accurately record, and the evidence related to lifestyle research is not yet clear. Our study is not strong enough to address this issue. The association can be studied in future research.

Strengths of this study included a school-based design which had a relatively large sample size and the availability of multivariate confounding factors adjusted in the model. The subgroup analyses further supported the robustness of the study findings. HVA more accurately reflects children's daily visual state and was included in the discussion of associated factors. In our study, SE and CYL anisometropia were considered, which the latter being disregarded in several studies. Several limitations should also be noted. Firstly, constrained by cross-sectional studies, no causal relationship between associated factors and anisometropia has been demonstrated. Lastly, the collection of outdoor activity time and near work time by self-reported questionnaires inevitably leads to recall bias.

In summary, we found that anisometropia in children aged 4 to 17 was likely the result of a combination of genetic and environmental factors. Preventing the occurrence of myopia and early treating anisometropic amblyopia might help to control the further development of anisometropia. Ophthalmologists and optometrists should pay attention to the changes of difference in inter-eye SE, AL, CA, and HVA, stereoacuity, to identify the current situation of anisometropia and the effect of the intervention. Indoor near work time and outdoor activity time may be associated with the occurrence of anisometropia.