sign in sign up
<<
Home , Diabetic retinopathy

is Inversely Associated With the Prevalence of Diabetic in the South Korean Population

DANIEL L. CHAO, SHUAI-CHUN LIN, REBECCA CHEN, AND SHAN C. LIN

PURPOSE: Diabetic retinopathy is a major cause of IABETIC RETINOPATHY IS A SIGHT-THREATENING irreversible vision loss. Recent studies have suggested microvascular of mellitus that myopia may be negatively correlated with the prev- (DM), and is the leading cause of blindness in D 1 alence of diabetic retinopathy. We sought to further the working age population worldwide. Numerous epide- investigate the association between and miologic studies have identified risk factors such as hyper- the likelihood of having diabetic retinopathy in a cross- tension, renal impairment, duration of diabetes, glycemic sectional, population-based study of the South Korean control, and use of as risk factors for development population. of diabetic retinopathy.2 Interestingly, both cross-sectional DESIGN: Cross-sectional study. clinic-based studies and large population surveys have sug- METHODS: Data were included from right eyes of gested a negative association between myopia and diabetic 13 424 participants who were 40 years and older with retinopathy, suggesting that myopia may have a protective gradable photographs of the Fourth and the effect on developing diabetic retinopathy. This was first sug- Fifth Korea National Health and Nutrition Examination gested in cross-sectional clinic-based studies in diabetic Survey. Diabetic retinopathy was graded using standard patients with similar HLA phenotypes, showing that fundus photographs. Autorefraction data were collected nonmyopic subjects were more likely to have nonprolifera- to calculate spherical equivalent of refraction in diopters tive diabetic retinopathy or proliferative diabetic retinop- (D) and further classified into 4 groups: hyperopia athy.3,4 An inverse association between myopia and (‡1.0 D), emmetropia (L0.99 D to 0.99 D), mild proliferative diabetic retinopathy was also found in the myopia (L1.0 D to L2.99 D), and moderate to high Wisconsin Epidemiologic Study of Diabetic Retinopathy myopia (£L3.0 D). Demographic, comorbidity, and (WESDR) study, where those with myopia <2 diopters health-related behavior information was obtained via (D) spherical equivalent (SE) were 60% less likely to interview. A multivariate model was used to evaluate progress to proliferative diabetic retinopathy (PDR) in the association between the diagnosis of any diabetic younger-onset diabetes.5 A multivariate analysis of the retinopathy and the refractive status. Singapore Malay Eye Study (SiMES) also demonstrated RESULTS: Mild myopia and moderate to high myopia that eyes with more severe myopia were less likely to have groups were negatively associated with development of any diabetic retinopathy.6 In contrast, the Beijing Eye Study any diabetic retinopathy (odds ratio [OR] 0.42; 95% did not demonstrate an association between myopia and confidence interval [CI] 0.18–0.97 and OR 0.14; 95% diabetic retinopathy in a univariate analysis.7 Thus, though CI 0.02–0.88, respectively). In addition, for every 1 D some studies suggest a negative association between myopia increase in spherical equivalent, there was a 30% increase and diabetic retinopathy, there are still conflicting data of having diabetic retinopathy (OR 1.30; 95% CI, among large population-based epidemiologic studies. 1.08–1.58). In this study we use the Korean National Health CONCLUSIONS: Our results from a population-based and Nutrition Examination Survey (KNHANES), a large study suggest that myopic status is associated with lower prospective population-based cross-sectional health study odds of having diabetic retinopathy in the South Korean that included for diabetic retinopathy, to assess population. (Am J Ophthalmol 2016;172:39–44. Ó the association between myopia and diabetic retinopathy 2016 Elsevier Inc. All rights reserved.) in this population. The findings may be particularly rele- vant in this population, which has a high rate of myopia and a growing population of diabetics.

Supplemental Material available at AJO.com. Accepted for publication Sep 7, 2016. From the Department of ; University of California, San Francisco, San Francisco, California. METHODS Daniel L. Chao is currently at the Shiley Eye Institute, UCSD Depart- ment of Ophthalmology, La Jolla, California. Inquiries to Shan C. Lin, 10 Koret Way, Room K301, San Francisco, THIS STUDY IS AN ANALYSIS OF A LARGE PROSPECTIVE CA 94143-0730; e-mail: [email protected] population-based cross-sectional health study of South

0002-9394/$36.00 © 2016 ELSEVIER INC.ALL RIGHTS RESERVED. 39 http://dx.doi.org/10.1016/j.ajo.2016.09.011 Koreans. The survey has been conducted annually since EVALUATION OF DIABETIC RETINOPATHY: Evaluation 2007 under the auspices of the Korea Centers for Disease of diabetic retinopathy in the KNHANES study was Control and Prevention, with approval by its institutional described previously.8 In participants with a history of review board. DM, random level >200 mg/dL, or a suspi- cious diabetic DR finding in nonmydriatic digital fundus STUDY POPULATION: All analyses were based on data photographs (TRC-NW6S; Topcon, Tokyo, Japan), which from the fourth and fifth KNHANES performed from July was performed in all participants >40 years old, 7 standard 2008 to December 2011. The KNHANES is a cross- photographs from the Early Treatment Diabetic Retinop- sectional survey that examines the health and nutritional athy Study (ETDRS) were obtained from both eyes after status of the noninstitutionalized civilian South Korean pharmacologic dilation. Retinopathy was identified population. The KNHANES consists of the health inter- if any characteristic lesion as defined by the ETDRS view, health behavior and nutrition surveys, and a health severity scale was present: microaneurysms (MAs), hemor- examination. The survey adhered to the principles outlined rhages, cotton-wool spots (CWSs), intraretinal microvas- in the Declaration of Helsinki for research that involves cular abnormalities (IRMAs), hard exudates (HEs), humans, and all participants provided written informed venous beading, and new vessels. A retinopathy severity consent. This nationwide representative study for noninsti- score was assigned to each eye according to the modifica- tutionalized civilians uses a stratified, multistage probabil- tion of the Airlie House Classification system as previously ity sampling design with a rolling survey sampling model. described.8,9 Eyes were graded according to the following Data on demographic characteristics, diet, and health- criteria: no DR (levels 10–13) or any DR (levels 14–80). related variables were collected through personal interview The DR was divided further into nonproliferative DR and a self-administered questionnaire. Physical examina- (NPDR, levels 14–60) and proliferative DR (level >60). tion and blood and urine sampling were performed at a mobile examination center. Ophthalmologic interview REFRACTIVE STATUS: The primary predictor variable questions and examinations were added in the second half was refractive error, which was evaluated by autorefrac- of 2008 and were thus available for KNHANES IV and V. tion (KR-8800; Topcon) without . Refraction In KNHANES, both the 1-year data surveys and the was converted to spherical equivalent, calculated as the integrated data of the 2008 through 2011 surveys repre- spherical value plus half of the astigmatic value. We cate- sent the entire population of Korea. Response rates were gorized refractive status into 4 groups: emmetropia (0.99 77.8%, 82.8%, 81.9%, and 80.4% in 2008, 2009, 2010, to 0.99 D), mild myopia (1.00 to 2.99 D), moderate to and 2011, respectively (9744 of 12 528 in 2008, 10 533 high myopia (<_3.00 D), and hyperopia (>_1.00 D). of 12 722 in 2009, 8958 of 10 938 in 2010, and 8518 of We then evaluated the odds of having any diabetic reti- 10 589 in 2011). There were 14 932 subjects 40 years of nopathy, nonproliferative diabetic retinopathy, and pro- age and older who participated during the 4-year study liferative diabetic retinopathy of each refractive group period with gradable fundus photographs. We further compared with the emmetropia group. excluded participants who had a history of refractive surgery and anyone who was aphakic or pseudophakic SLIT-LAMP EXAMINATION: We also investigated the (n ¼ 1087), or those in which no refraction data were relationship between diabetic retinopathy and anterior available (n ¼ 421). This left 13 424 subjects who were chamber depth (ACD) and status. A slit-lamp then included in the study. examination (Haag-Streit model BQ-900; Haag-Streit AG, Koeniz, Switzerland) was performed by study ophthal- SURVEY COMPONENTS: The methods related to mologists for determination of diseases in the anterior the ophthalmologic examination in the KNHANES segment of the eye (eg, , cataract, , and have been described in prior publications.8 After an pseudophakia) and measurement of the IOP and the ophthalmology-focused interview, participants underwent ACD using the Van Herick method.8 Cataract status was measurements, automated refraction, slit- defined as a nuclear, cortical, anterior subcapsular, poste- lamp examination, intraocular pressure (IOP) measure- rior subcapsular, and mixed-type cataract. Pseudophakic ment, , and, when deemed appropriate, and aphakic eyes were excluded from our analyses. visual field (VF) examination. Potential confounding variables that were considered included age; sex; income status; educational level; EVALUATION OF DIABETES MELLITUS: Those who were health-related behavior, such as smoking, alcohol use, diagnosed by a self-reported history of a physician diagnosis exercise, and body mass index; and medical comorbidities, or those who were receiving drug treatment for DM, such as anemia, renal failure, hypertension, hyperlipid- including insulin or oral hypoglycemic agents, and those emia, angina, and . In addition to assessment of de- who had a fasting plasma glucose level >126 mg/dL mographic information and medical history, participants without a previous diagnosis of DM were classified as underwent examination of and testing of subjects with DM. blood and urine.

40 AMERICAN JOURNAL OF OPHTHALMOLOGY DECEMBER 2016 STATISTICAL ANALYSIS: Complex sample analysis was there was a 1.3 increase in the odds ratio of diabetic reti- used for the KNHANES IV and V data for weighting nopathy (OR 1.3; 95% CI 1.08–1.58). all values following statistical guidance from the Korea We further looked to see whether myopia was associated Centers for Disease Control and Prevention. The regres- with particular subsets of diabetic retinopathy such as sion model was constructed after identification of potential NPDR or PDR. Multivariate analyses adjusting for the confounding variables. All risk factors that were identified same variables looking for an association between myopia as being associated with any DR, NPDR, and PDR diag- and PDR did not show any significant association (OR nosis by univariate analysis with P < .1 as the cutoff point 1.02; 95% CI 0.77–1.35), while high myopia (<_6.0 D) were then included in the multivariable analysis to assess was negatively associated with NPDR (OR 0.01 95% CI the possible independent association between diabetic reti- 0.001–0.22) (Table 3). In addition, for every 1 diopter in- nopathy and the refractive status. After ascertainment of crease in spherical equivalent, there was an increased odds such a possible association, 95% confidence intervals ratio of 1.3 of having NPDR (OR 1.33, 95% CI 1.08–1.64). [CIs] of odds ratios (ORs) were identified for each possible association. Two-sided statistical tests were performed with SPSS statistical software, version 21.0 (IBM, Armonk, New York, USA). DISCUSSION

IN THIS STUDY, WE DEMONSTRATE A NEGATIVE ASSOCIA- tion between myopia and the prevalence of diabetic reti- RESULTS nopathy in the KNHANES population. This is consistent with previous epidemiologic studies of myopia and diabetic THERE WERE 14 932 SUBJECTS 40 YEARS OF AGE AND OLDER retinopathy showing a reduced incidence of diabetic reti- who participated during the 4-year study period with grad- nopathy in myopic patients.10,11 Interestingly, a previous able fundus photographs. We excluded 1087 subjects with analysis for risk factors of diabetic retinopathy in the history of refractive surgery or who had aphakia or pseu- KNHANES population did not find an association dophakia and 421 subjects with no refraction data between myopia or hyperopia and diabetic retinopathy.9 OD, leaving 13 424 eligible participants for our analyses They measured the presence or absence of either myopia (Table 1). The SE was highly correlated between eyes: or hyperopia (defined as either <1.0 D SE or >1.0 D the mean SE was 0.50 (6 0.02) D in the right eye SE, respectively) and its association with diabetic retinop- and 0.47 (6 0.02) D in the left eye (r ¼ 0.867, athy. We repeated this analysis using <1.0 D SE as a cut- P < .001), so we therefore used the right eye for all further off in our multivariate model for the right eye and also did analyses on refractive error. not find a statistically significant association (OR 0.74; We collected data from right eyes of 13 424 subjects, of 95% CI 0.48–1.14). We suggest that this definition of whom 1685 had a diagnosis of diabetes mellitus. Of those myopia may have been too broad, and thus did not reveal 1685 subjects, 168 subjects had diabetic retinopathy this association as when we further substratified the myopia (9.8%), with 147 subjects (8.7%) classified as NPDR groups. In addition, the lack of exclusion of patients who and 21 (1.1%) classified as PDR. Table 1 presents a com- had cataract surgery, were aphakic, or had refractive surgery parison of demographic data between diabetic patients may also have obscured these results. Interestingly, the with and without diabetic retinopathy. Demographic overall prevalence of diabetic retinopathy (9.8%) is lower characteristics, comorbidities, health-related behavior, than reported in another epidemiologic studies. Previous and eye conditions such as refractive status, ACD, and studies of the prevalence of diabetic retinopathy of Asian the cataract status among subjects with different diagno- populations shows a range of prevalence of diabetic reti- ses are described. nopathy from 16% to 35%.2,7,9 Whether our lower We next assessed the relationship between refractive prevalence of diabetic retinopathy is attributable to error and presence of diabetic retinopathy in patients sampling bias, differing criteria for characterizing diabetic with diabetes mellitus. A multivariate logistic regression retinopathy, or a difference between South Korean and analysis adjusting for age, sex, income, education, cataract other Asian populations in the prevalence of diabetic subtypes, exercise, body mass index, fasting blood sugar, retinopathy remains to be determined. HbA1c, and BUN was performed looking at the association This negative association between myopia and diabetic between refractive status and diabetic retinopathy in sub- retinopathy has been noted in many studies; however, jects with diabetes mellitus. We found that both the mild there have also been conflicting results.2,5–7,12,13 These myopia (OR 0.42; 95% CI 0.18–0.97) and the moderate studies have included population-based cross-sectional to high myopia groups (OR 0.14; 95% CI 0.02–0.88) had studies as well as longitudinal studies encompassing significantly negative associations with the diagnosis of dia- American, Australian, and Asian populations. This may betic retinopathy (Table 2). In addition, modeling showed be due to inconsistent definitions or grouping of myopia, that for every 1 diopter increase in spherical equivalent, as well as whether refractive myopia (SE) or axial myopia

VOL. 172 MYOPIA AND DIABETIC RETINOPATHY 41 TABLE 1. Demographic Characteristics of Diabetic Patients TABLE 1. Demographic Characteristics of Diabetic Patients With and Without Diabetic Retinopathy With and Without Diabetic Retinopathy (Continued)

No DR (N ¼ 1517, DR (N ¼ 168, No DR (N ¼ 1517, DR (N ¼ 168, Characteristic 90.2%) 9.8%) Characteristic 90.2%) 9.8%)

Eye conditions Medical comorbidities Refractive status 0.31 (0.07) 0.11 (0.13) Anemia 145 (9.6) 22 (10.8) (SE, diopters) Renal failure 10 (0.5) 0 (0) Hyperopia (SE >_1.0) 414 (20.4) 40 (22.6) MI 35 (1.7) 4 (2.4) Emmetropia 818 (57.1) 103 (65.6) Hypertension 798 (49.0) 83 (41.7) (0.99 <_ SE <_ 0.99) Hyperlipidemia 379 (22.8) 40 (21.7) Mild myopia 202 (15.7) 19 (9.0) Angina 62 (3.4) 9 (4.3) (2.99 < SE <_ 1.0) Stroke 72 (3.6) 10 (7.7) Moderate to high myopia 83 (6.8) 6 (2.8) BUN 15.72 16.76 (SE <_3.0) Creatine 0.87 0.88 a ACD Triglycerides 188 193 <_1/4 29 (1.8) 2 (1.4) Mean SBP 126 129 1/4 < ACD <1/2 395 (25.9) 42 (28.4) Mean SDP 79 77 >_1/2 1071 (72.3) 122 (70.1) Blood sugar 141 171 Cataract HbA1c 7.17 8.52 Cortical 206 (11.6) 21 (9.5) Nuclear 513 (30.0) 64 (34.7) ACD ¼ anterior chamber depth; BMI ¼ body mass index; Anterior subcapsular 23 (1.0) 1 (1.0) BUN ¼ blood urea nitrogen; D ¼ diopters; DBP ¼ diastolic blood Posterior subcapsular 8 (0.5) 3 (3.6%) pressure; DR ¼ diabetic retinopathy; SBP ¼ systolic blood pres- Mixed 159 (7.7) 25 (14.9) sure; SE ¼ spherical equivalent (sphere [D] þ 1/2 cylinder [D]). No cataract 608 (49.3) 54 (36.4) Data are presented as number (percentage) of study partici- Demographics pants unless otherwise indicated. Percentages may not total Age, mean (SE), y 58.37 (0.37) 59.33 (0.97) 100 because of missing values in the database. a Women, n (%) 719 (42.2) 85 (45) ACD was accessed using the Van Herick method. b Education Calculated as weight in kilograms divided by height in meters Elementary and under 694 (38.9) 84 (47.4) squared. Middle school 239 (16.3) 32 (21.7) High school 361 (29.3) 32 (17.6) College and graduate 201 (15.6) 16 (13.2) Annual income, quartile 1st 482 (26.8) 61 (38.9) 2nd 377 (27.5) 38 (21.5) 3rd 322 (22.5) 37 (22.8) 4th 307 (23.1) 28 (16.8) TABLE 2. Health-related behavior Multivariate Logistic Regression Analysis Looking at Association of Refractive Error to Presence of Diabetic Duration of diabetes, mean in 6.85 (0.25) 10.75 (0.83) Retinopathy years (standard error) Smoking status 95% 95% Never 792 (47.8) 94 (54.4) OD SE OR CI - Lower CI - Higher Past smokers 210 (13.3) 23 (13.2) Hyperopia (>_1.0 D) 1.13 0.60 2.12 Current smokers 515 (38.9) 51 (32.4) Emmetropia 1.00 1.00 1.00 Alcohol (0.99 D to 0.99 D) No drinking in past year 256 (18.5) 34 (30.3) Mild myopia 0.42 0.18 .97 <_1/month 220 (17.8) 21 (12.7) (1.0 D to 2.99 D) 2–4 times/month 219 (20.3) 21 (21.3) Moderate to high myopia 0.14 0.02 .88 2–3 times/week 190 (17.8) 16 (13.5) (<_3.0 D) >_ 4/week 164 (15.6) 20 (13.6) b Per 1 D SE 1.30 1.08 1.58 BMI, mean (SE) 25.19 (0.11) 23.82 (0.25) Exercise (days per week) CI ¼ confidence interval; D ¼ diopter; OD ¼ right eye; 0 1103 (68.7) 140 (85.2) OR ¼ odds ratio; SE¼ spherical equivalent. 1 86 (6.8) 3 (3.3) Adjusted for age, sex, household income, education, cataract 2 83 (6.6) 4 (1.7) subtypes, exercise, body mass index, fasting blood sugar, 3 91 (7.8) 8 (4.0) HbA1c, blood urea nitrogen. >_4 109 (7.2) 9 (5.4)

42 AMERICAN JOURNAL OF OPHTHALMOLOGY DECEMBER 2016 measured, such as the Beijing Eye Study and SiMES study. TABLE 3. Multivariate Logistic Regression Analysis Looking The Beijing Eye Study did not show an association between at Association of Refractive Error to Presence of ACD and diabetic retinopathy after adjustment for multi- Nonproliferative Diabetic Retinopathy ple confounding variables.7 However, the SiMES did show that larger ACD was associated with decreased inci- 95% 95% 2 OD SE OR CI - Lower CI - Higher dence of moderate or vision-threatening DR. Additional studies are needed to determine whether ACD is associated Hyperopia (>_1.0 D) 1.25 0.65 2.40 with diabetic retinopathy. Emmetropia 1.00 1.00 1.00 (0.99 D to 0.99 D) The pathophysiology of how myopia may be protective Mild myopia 0.38 0.14 1.03 in diabetic retinopathy is still controversial. Many hy- (1.0 D to 2.99 D) potheses have been put forward, mostly focused on axial Moderate myopia 0.46 0.09 2.34 elongation of the eye. One hypothesis is that with axial (3.0 D to 5.99 D) elongation, there is narrowing of blood vessels in the High myopia (<_6.0 D) 0.01 0.001 0.22 retina leading to reduction in retinal blood flow according Per 1 D myopia 1.33 1.08 1.64 to the Hagen-Poiseuille law,16 resulting in a decrease in capillary hydrostatic pressure and consequent decreased CI ¼ confidence interval; D ¼ diopter; OD ¼ right eye; ¼ ¼ likelihood of leakage and rupture of compromised retinal OR odds ratio; SE spherical equivalent. 17 Adjusted for age, sex, household income, education, cataract capillaries in diabetes (Starling and Laplace’s law). subtypes, exercise, body mass index, fasting blood sugar, Increased retinal blood flow has been linked to progres- HbA1c, blood urea nitrogen. sion of DR, and normal myopic eyes have decreased retinal blood flow compared with emmetropic eyes.18,19 However, a recent study suggested that retinal capillary (axial length) were measured. However, when taken in blood flow is not related to axial length or incidence of aggregrate, 2 separate meta-analyses looking at the associ- diabetic retinopathy.20 ation between myopia and diabetic retinopathy both show A second hypothesis is that elongated eyes have thinner a negative association between myopia and diabetic reti- , which results in decreased metabolic demand in nopathy.10,11 the . The decreased metabolic demand then could Whether it is the refractive (ie, corneal curvature, spher- blunt the hypoxic response that is necessary for diabetic ical equivalent) or structural (ie, axial length) component retinopathy. A recent study of healthy myopic eyes demon- of myopia that leads to this negative association with strated decreased O2 consumption and decreased retinal diabetic retinopathy is still up for debate. Many of the function by multifocal in myopic population-based studies, including KNHANES, did not eyes.21 The increased ocular volume owing to the eye elon- measure axial length. Based on a number of other studies, gation may also play a factor as well, diluting the levels of axial length appears to be strongly correlated with angiogenic cytokines such as vascular endothelial growth decreased incidence of diabetic retinopathy. A cross- factor (VEGF). Indeed, a negative correlation between sectional study of diabetic retinopathy showed that axial aqueous VEGF levels and axial length has been length, and not corneal curvature or spherical equivalent, described.22 A third hypothesis is that myopia is associated was negatively associated with both diabetic retinopathy with increased incidence of posterior vitreous detach- and diabetic (DME) in a dose-dependent ments, which have been found to be protective in diabetic manner.14 Additionally, an analysis of studies also sug- retinopathy. gested that axial length, and not refractive myopia, is the Examining the relationship between myopia and dia- principal component of the negative association of myopia betic retinopathy is of particular relevance in the Korean with diabetic retinopathy.15 Indeed, both recent meta- population, given the increasing prevalence of both myopia analyses demonstrated a strong association between and diabetes in the Korean population. In a recent cross- increasing axial length and decreased incidence of diabetic sectional survey, 76.4% of Koreans aged 19–29 were retinopathy.10,11 Thus, axial myopia is likely a strong factor myopic (<0.5 D), compared with 16% in >70-year-olds. in protecting from diabetic retinopathy, and other This is also true for high myopia (>6 D), with 10.1% of refractive components still need to be studied for their 19- to 29-year-olds highly myopic compared with only association with diabetic retinopathy. 1.1% of 60-year-olds. The prevalence of diabetes has also One component of axial myopia is ACD. In our study we risen 6- to 7-fold in Korea, from 1.9% to 9.9%, in the last did not demonstrate an association between ACD and 40 years.2 Given the high and increasing prevalence of diabetic retinopathy (P ¼ .867), suggesting that the vitre- both of these conditions, understanding the relationship ous chamber dimension may be more important to this between myopia and diabetic retinopathy may have impor- effect. It should be noted that ACD was measured qualita- tant public health implications. tively using a Van Herrick method and thus should not be Our study has some limitations. First, although the compared to other studies where absolute ACD was association of myopia and diabetic retinopathy has

VOL. 172 MYOPIA AND DIABETIC RETINOPATHY 43 been assessed, we did not evaluate every component study, this association still needs to be confirmed thro- associated with refractive status, such as axial length ugh future studies. and corneal curvature, given that such data were not In summary, we found that after adjustment for potential available in the KNHANES. Second, cross-sectional confounding variables, myopic status was associated with population studies do not allow for the determination significantly lower odds of any diabetic retinopathy in a of causation. This work simply reports an association large population-based sample of South Koreans aged 40 between the refractive status and the prevalence of years and older. Future prospective and laboratory studies diabetic retinopathy. Although myopia seems to have are needed to provide a clearer understanding of the rela- some protective effect on diabetic retinopathy in our tionship between myopia and diabetic retinopathy.

FUNDING/SUPPORT: THIS STUDY WAS SUPPORTED BY CORE GRANT EY002162 FROM THE , BETHESDA, Maryland (S.-C.L.); by That Man May See, Inc, San Francisco, California (S.-C.L.); and by Research to Prevent Blindness, New York, New York (S.-C.L.). Financial disclosures: Daniel L. Chao: Consultant, Allergan. The following authors have no financial disclosures: Shuai-Chun Lin, Rebecca Chen, and Shan C. Lin. All authors attest that they meet the current ICMJE criteria for authorship.

REFERENCES 12. Pan CW, Cheung CY, Aung T, et al. Differential associa- tions of myopia with major age-related eye diseases: the 1. Klein BE. Overview of epidemiologic studies of diabetic Singapore Indian Eye Study. Ophthalmology 2013;120(2): retinopathy. Ophthalmic Epidemiol 2007;14(4):179–183. 284–291. 2. Wong TY, Cheung N, Tay WT, et al. Prevalence and risk 13. Yang KJ, Sun CC, Ku WC, et al. Axial length and factors for diabetic retinopathy: the Singapore Malay Eye proliferative diabetic retinopathy. Optom Vis Sci 2012; Study. Ophthalmology 2008;115(11):1869–1875. 89(4):465–470. 3. Rand LI, Krolewski AS, Aiello LM, Warram JH, Baker RS, 14. Man RE, Sasongko MB, Sanmugasundram S, et al. Longer Maki T. Multiple factors in the prediction of risk of prolifer- axial length is protective of diabetic retinopathy and macular ative diabetic retinopathy. N Engl J Med 1985;313(23): edema. Ophthalmology 2012;119(9):1754–1759. 1433–1438. 15. Man RE, Sasongko MB, Wang JJ, Lamoureux EL. Association 4. Baker RS, Rand LI, Krolewski AS, Maki T, Warram JH, between myopia and diabetic retinopathy: a review of obser- Aiello LM. Influence of HLA-DR phenotype and myopia vational findings and potential mechanisms. Clin Experiment on the risk of nonproliferative and proliferative diabetic Ophthalmol 2013;41(3):293–301. retinopathy. Am J Ophthalmol 1986;102(6):693–700. 16. Delaey C, Van De Voorde J. Regulatory mechanisms in the 5. Moss SE, Klein R, Klein BE. Ocular factors in the incidence retinal and choroidal circulation. Ophthalmic Res 2000; and progression of diabetic retinopathy. Ophthalmology 32(6):249–256. 1994;101(1):77–83. 17. Quigley M, Cohen S. A new pressure attenuation index to 6. Lim LS, Lamoureux E, Saw SM, Tay WT, Mitchell P, evaluate retinal circulation. A link to protective factors in Wong TY. Are myopic eyes less likely to have diabetic diabetic retinopathy. Arch Ophthalmol 1999;117(1):84–89. retinopathy? Ophthalmology 2010;117(3):524–530. 18. Dimitrova G, Kato S, Yamashita H, et al. Relation between 7. Xie XW, Xu L, Wang YX, Jonas JB. Prevalence and associated retrobulbar circulation and progression of diabetic retinop- factors of diabetic retinopathy. The Beijing Eye Study 2006. athy. Br J Ophthalmol 2003;87(5):622–625. Graefes Arch Clin Exp Ophthalmol 2008;246(11):1519–1526. 19. Shimada N, Ohno-Matsui K, Harino S, et al. Reduction of 8. Yoon KC, Choi W, Lee HS, et al. An overview of ophthal- retinal blood flow in high myopia. Graefes Arch Clin Exp mologic survey methodology in the 2008-2015 Korean Na- Ophthalmol 2004;242(4):284–288. tional Health and Nutrition Examination Surveys. Korean 20. Man RE, Sasongko MB, Xie J, et al. Decreased retinal J Ophthalmol 2015;29(6):359–367. capillary flow is not a mediator of the protective myopia- 9. Jee D, Lee WK, Kang S. Prevalence and risk factors for dia- diabetic retinopathy relationship. Invest Ophthalmol Vis Sci betic retinopathy: the Korea National Health and Nutrition 2014;55(10):6901–6907. Examination Survey 2008-2011. Invest Ophthalmol Vis Sci 21. Man RE, Lamoureux EL, Taouk Y, et al. Axial length, retinal 2013;54(10):6827–6833. function, and oxygen consumption: a potential mechanism 10. Fu Y, Geng D, Liu H, Che H. Myopia and/or longer axial for a lower risk of diabetic retinopathy in longer eyes. Invest length are protective against diabetic retinopathy: a meta- Ophthalmol Vis Sci 2013;54(12):7691–7698. analysis. Acta Ophthalmol 2015;94(4):346–352. 22. Sawada O, Miyake T, Kakinoki M, Sawada T, Kawamura H, 11. Wang X, Tang L, Gao L, Yang Y, Cao D, Li Y. Myopia and Ohji M. Negative correlation between aqueous vascular diabetic retinopathy: a systematic review and meta-analysis. endothelial growth factor levels and axial length. Jpn J Diabetes Res Clin Pract 2016;111:1–9. Ophthalmol 2011;55(4):401–404.

44 AMERICAN JOURNAL OF OPHTHALMOLOGY DECEMBER 2016