We used NHANES 2005–2008 data, which was collected from a public source with a complex multistage survey design for our cross-sectional study, and the results were consistent in different statistical models. After adjusting all the factors, we substantiated that the protective effect of high myopia was associated with a reduced likelihood of diabetic retinopathy (OR: 0.44, 95% CI: 0.18–0.96) In the stratified analyses, subjects with high myopia levels did have a significantly lower risk of diabetic retinopathy in the non-Hispanic Black group (OR: 0.20, 95%CI: 0.04–0.95).

The prevalence of people with high myopia from this screening data was 5.69%, which is lower than the U.K. population (9.5%), but higher than the Chinese (3.93%) and elder Indian population (< 1%) [16,17,18]. Many epidemiological surveys had proposed that an increase in refractive is a protective influence against diabetic retinopathy. However, some studies showed conflicting results in severe myopia in meta-analysis studies [19,20,21,22].

In recent years, axial length has been considered to be one of the most important eye measurement standards for diagnosing myopia, and more than 26 mm can be regarded as high myopia [23, 24]. Therefore, the relationship between severe myopia and diabetic retinopathy was also assessed based on the axial length in some studies. Two Singaporean cohort studies had described that the OR value of subjects with diabetic retinopathy were 0.86 and 0.68 per 1 mm increase in axis length, respectively [7, 9]. As for the Chinese population in Beijing, Xu et al. explored the incidence of diabetic retinopathy and its related factors in 2602 participants during a ten-year study. The results showed a total of 109 new patients with diabetic retinopathy during the ten years. These patients had a shorter eye axis length than the non-diseased person (OR: 0.48; 95% CI: 0.33–0.71) [25]. The prevalence of diabetic retinopathy was not related to refractive measurements, but the long axial length was associated with lower diabetic retinopathy [12]. This conclusion was similar to another study of Chinese populations [26]. In the cross-section study, Wang et al. measured axial lengths as an eye parameter for the prevalence and severity of diabetic retinopathy. A higher prevalence of diabetic retinopathy is associated with shorter axial lengths (OR: 0.81, 95% CI: 0.70–0.95) [13]. Unfortunately, although longer axial length is closely related to diabetic retinopathy, the structural component of participants was not measured in the NHANES datasets. Thus, we could not discuss the influence of axial length in our analysis even if it's an important diagnostic indicator.

According to the above-mentioned, Myopia is widely considered to be one of the protective factors for DR, but most studies have explored the situation in Asia. Few studies have explored the relationship between high myopia and DR in different Western populations based on large sample sizes. We found that high myopia has a significant negative relationship with DR in black people. There was a negative correlation between high myopia interval and diabetic retinopathy in univariate analysis (OR: 0.39, 95% CI: 0.18–0.84). After adjustment for age, gender, and race (Model II), high myopia might reduce the risk of diabetic retinopathy compared with those in the emmetropia group (OR: 0.43, 95% CI: 0.20–0.93). After further adjustment for education level, BMI, PIR, CVD, and smoking status (Model III), high myopia was still associated with a reduced diabetic retinopathy risk (OR: 0.44, 95% CI: 0.20–0.96).

To the best of our knowledge, there are no large-scale population-based studies that have focused on the relationship between high myopia and diabetic retinopathy in Western countries. Therefore, we investigated NHANES, a database with a complex sampling design in the United States population, and found high myopia is regarded as a factor influencing diabetic retinopathy in this public database. Moreover, the correlation between different races and diabetic retinopathy may be an influence factor. Nwanyanwu et al. described the relationship between race and the prevalence of diabetic retinopathy [27]. In the present study, High myopia is a protective factor for diabetic retinopathy in black populations.

Some limitations of this study should be acknowledged. First of all, the analysis was of cross-sectional design, thus our study cannot determine a causal relationship between two ocular diseases. Second, given the absence of data on the axial length of the eye in the NHANES dataset, we could not assess this main factor affecting refractive error. However, according to previous studies, some population-based studies also classify the degree of myopia solely based on SE value. If axial length assessment could be provided in later NHANES studies, combined with the SE classification method, it would provide more accurate measurements for myopia severity. Last, we had not analyzed the association between high myopia and proliferative diabetic retinopathy, which represents the last stage of this extremely complex retinal disease. This is because very few participants simultaneously suffered from high myopia and proliferative diabetic retinopathy in our analysis.