Prevalence and Risk Factors for Diabetic Retinopathy: The Korea National Health and Nutrition Examination Survey 2008–2011

Abstract

Purpose.: We evaluated the prevalence and risk factors for diabetic retinopathy (DR) in a representative Korean population.

Methods.: This cross-sectional study involved 16,109 subjects aged >40 years who had participated in the Korean National Health and Nutrition Survey from 2008 to 2011. Seven standard retinal fundus photographs were obtained after pupil dilatation from both eyes. The DR was graded using the modified Airlie House classification system. Risk factors for DR and vision-threatening diabetic retinopathy (VTDR) were evaluated, including age, sex, diabetes duration, glycated hemoglobin (HbA1c), hypertension, lipid profiles, and refractive error.

Results.: Of the 16,109 eligible individuals participating in the study, 14,595 (90.6%) had fasting blood glucose results available. Of these, 2023 (13.8%) were diagnosed with diabetes mellitus. Among these, gradable photographs were available for 1678 subjects (82.9%), including 1323 subjects with known diabetes mellitus (KDM) and 355 with newly diagnosed diabetes mellitus (NDM). The prevalences of any DR and VTDR were 15.8% (95% confidence interval [CI], 14.1–17.5), and 4.6% (95% CI, 3.6–5.6), respectively. Any DR was associated with a longer duration of diabetes (odds ratio [OR], 1.08; 95% CI, 1.06–1.10), higher HbA1C level (OR, 1.52; 95% CI, 1.28–1.80), higher systolic blood pressure (OR, 1.02; 95% CI, 1.01–1.03), and lower body mass index (OR, 0.91; 95% CI, 0.87–0.96) in a multivariate analysis.

Conclusions.: We provide data on the prevalence and risk factors for DR in a representative Korean population. The prevalence of any DR and VTDR in the Korean population was lower than that reported previously. The condition of DR was associated with a longer duration of diabetes, poor glycemic control, and higher systemic blood pressure.

Introduction

Diabetic retinopathy (DR) is the leading cause of blindness among working-aged adults worldwide. ¹ It is estimated that the number of people with diabetes mellitus (DM) worldwide will double by 2025, resulting in approximately 300 million people with this condition. ² A rising prevalence of DM has been identified in emerging Asian countries, such as India and China. ^3,4 The prevalence of DM in Korea has increased 6- to 7-fold from 1.5% to 9.9% in the past 40 years. ⁵ The increased prevalence of DM in Korea likely is attributable to rapid economic development, improved living standards, an aging population, and a westernized lifestyle. Thus, DM and its complications are becoming a major public health problem in Korea. However, data on the prevalence and severity of DR in Korea are lacking. ^6–8 To our knowledge, only one population-based report of the prevalence of DR in a rural area of Korea is available. ⁸ In that study, indirect funduscopy was used, instead of fundus photography, to assess DR and no information on macular edema (ME) or vision-threatening DR (VTDR) was provided. 

The clinical characteristics of DM in Korea are different from those of western countries. ^5,9,10 The incidence of type 1 diabetes in Korea is one of the lowest worldwide. ⁵ A significant portion of diabetic patients are nonobese with type 2 DM, and their body weight often decreases during the disease course. ⁹ In addition, impaired insulin secretion may have a role in the pathogenesis of nonobese type 2 DM in Korea. ¹⁰ Thus, the prevalence and characteristics of DR in Korea may be different from those of western countries. 

Although many epidemiologic studies have assessed various risk factors for DR, the consistency and strength of these risk factors varied among studies. ^1,11,12 A possible reason for such inconsistencies may reside in insufficient numbers of study participants. ¹¹ Therefore, modifiable risk factors must be determined in studies with sufficiently large numbers of participants. 

Generating a more precise estimate of the prevalence of DR and its relationship with major modifiable risk factors is crucial for guiding public health education and optimal clinical management of DR. Therefore, we examined the prevalence of DR and evaluated its associated risk factors in a representative Korean population aged >40 years, using a stratified, multistage sampling method. 

Subjects and Methods

Study Population

This study was based on data acquired in the Korea National Health and Nutrition Survey (KNHNS), 2008 to 2011. The KNHNS is an ongoing, population-based, cross-sectional, and nationally representative survey conducted by the Division of Chronic Disease Surveillance, Korean Center for Disease Control and Prevention. The survey consisted of a health interview, a nutritional survey, and a health examination survey. The survey collected data via household interviews and by direct standardized physical examinations conducted in a specially-equipped mobile examination center. The sample design and size are estimated by KNHNS, so the results can be generalized to the entire Korean population. 

Annually, 4000 households in 200 enumeration districts were selected by a panel to represent the civilian, noninstitutionalized South Korean population using the systemic stratified, multistage clustered sampling method based on the National Census Data. All members of each selected household were asked to participate in the survey, and a participation rate between 2008 and 2011 ranged from 77.8% to 82.8%. 

All participants provided informed consent. This study design followed the tenets of the Declaration of Helsinki for biomedical research and was approved by the Institutional Review Board of the Catholic University of Korea, Seoul, Korea. 

Data Collection

Participants were considered to have type 1 DM if they were aged <30 years when diagnosed with DM and were receiving insulin therapy. ¹³ Otherwise, DM was considered type 2. Those who were diagnosed by a self-reported history of a physician diagnosis or those who were receiving drug treatment for DM, including insulin or oral hypoglycemic agents, were classified as subjects with known diabetes mellitus (KDM), while those who had a fasting plasma glucose level > 126 mg/dL without a previous diagnosis of DM were classified as subjects with new diabetes mellitus (NDM). ¹³  

In participants with a history of DM, random blood glucose level > 200 mg/dL, or a suspicious diabetic DR finding in nonmydriatic 45° digital fundus photographs (TRC-NW6S; Topcon, Tokyo, Japan), which was performed in all participants > 40 years old, seven standard photographs from the Early Treatment for Diabetic Retinopathy Study (ETDRS) were obtained from both eyes after pharmacologic pupil dilatation. 

Retinopathy was identified if any characteristic lesion as defined by the ETDRS severity scale was present: microaneurysms (MAs), hemorrhages, cotton wool spots (CWSs), intraretinal microvascular abnormalities (IRMAs), hard exudates (HEs), venous beading, and new vessels. A retinopathy severity score was assigned to each eye according to the modification of the Airlie House Classification system as follows ¹⁴ : Level 10, no retinopathy present; level 14, any combination of definite HE, CWS, IRMA, or venous loops in the absence of definite MA; level 15, hemorrhage present without any definite MA; level 20, MA only with no other diabetic lesion present; level 31, MA and one or more of the following: hemorrhage or MA less than standard photograph 2A, HE, venous loops, questionable CWS, IRMA, or venous beading; level 41, MA and one or more of the following: CWS, IRMA less than standard photograph 8A; level 51, MA and one or more of the following: venous beading, hemorrhage or MA > 2A, IRMA > 8A; level 60, fibrous proliferation with no other proliferative lesions; levels 61 to 64, laser scatter photocoagulation scars with retinopathy levels 31 to 51; level 65, proliferative DR less than high-risk characteristics, as defined in the Diabetic Retinopathy Study; level 70, proliferative DR with more than high risk characteristics; and level 80, total vitreous hemorrhage. 

The level of retinopathy was graded based on the worse eye. Eyes were graded according to the following criteria: no DR (levels 10–13) or any DR (levels 14–80). The DR was divided further into minimal nonproliferative DR (NPDR, levels 14–20), mild NPDR (level 31), moderate NPDR (levels 41–47), severe NPDR (level 51), and proliferative DR (level > 60). 

The condition of ME was defined by HE in the presence of MA and blot hemorrhage within one disk diameter from the foveal center, or the presence of focal photocoagulation scars in the macular area. Clinically significant ME (CSME) was considered present when the ME involved or was within 500 μm of the foveal center, or if focal photocoagulation scars were present in the ME. The VTDR was defined as the presence of severe NPDR, proliferative retinopathy, or CSME. ¹⁴  

Various associated factors were examined. Blood pressure (BP) was measured by standard methods using a sphygmomanometer with the patient in a sitting position. Three measurements were made in all subjects at 5-minute intervals, and the average of the second and third measurements was used in the analysis. Blood samples were collected in the morning after fasting for at least 8 hours. Fasting glucose, glycated hemoglobin (HbA1c), total cholesterol, triglycerides, low-density lipoprotein, cholesterol, and vitamin D levels were measured at a central, certified laboratory. The presence of DM was defined as fasting glucose > 126 mg/dL or if the individual was prescribed antiglycemic medication. The presence of hypertension was defined as a systolic BP >140 mm Hg or diastolic BP > 90 mm Hg, or if the individual was prescribed antihypertensive medication. Hypercholesterolemia was defined as a total cholesterol concentration > 240 mg/dL or if the individual was prescribed anticholesterol medication. Heart problems were defined as a history of myocardial infarction (MI) or angina. Refractive errors was defined as myopia (<−1.0 diopters [D]) and hyperopia (>+1.0 D). Body mass index (BMI) was calculated by dividing body weight by height squared (kg/m²) after height and weight were measured using standardized techniques and equipment. Obesity was defined as a BMI > 25 kg/m². Information regarding demographic and social factors was obtained using a standardized questionnaire during a health interview. 

Statistical Analyses

The age- and sex-specific prevalence of DR was assessed with 95% confidence intervals (CI). An ANOVA or the χ² test was used to compare the demographic characteristics. Logistic regression models were used to determine the risk factors for DR and VTDR. To test linearity for this logistic regression analysis, continuous variables were transformed into categoric variables, and odds ratios (ORs) were calculated for each transformed categoric variables. We identified the assumption of linear trend by gradual increase or decrease of ORs. The duration of DM was calculated as the difference between the year of diagnosis and the examination year of KNHNS. Those with NDM were given a DM duration of zero years. Sampling was weighted by statisticians by adjusting the oversampling and nonresponse rate. ¹⁵ Analyses were performed using the Statistical Package for the Social Sciences ver. 18.0 (SPSS, Inc., Chicago, IL). 

Results

Of the 16,109 eligible participants, 14,595 (90.6%) completed the fasting plasma glucose examination. Among the subjects with fasting plasma glucose results, 2023 (13.8%) subjects met the criteria for DM. Of these, 1678 (82.9%) participants had seven gradable standard photographs. 

The characteristics of the study population are shown in Table 1. Compared to nondiabetic subjects, patients with DM were more likely to be older, male, and have higher serum glucose, higher systolic BP, and higher total cholesterol, but less likely to drink alcohol. Among participants with DM, 6.7% reported previous angina or MI and 5.8% reported a previous stroke. 

Table 1

 

Characteristics of Participants in the Korean National Health and Nutrition Examination Survey, 2008 to 2011

Characteristics	No Diabetes, n = 12,572	Diabetes, n = 2023	P	Participants, n = 14,595	Nonparticipants, n = 1514	P
Age, per y	57.2 ± 11.6	63.1 ± 10.2	<0.001	58.0 ± 11.6	64.1 ± 13.8	<0.001
Sex, male	42.5	50.9	<0.001	43.7	38.3	<0.001
Fasting glucose, mg/dL	94.5 ± 9.6	141.2 ± 45.5	<0.001	101.0 ± 25.0	NA	NA
HbA1c, %	5.7 ± 0.4	7.3 ± 1.4	<0.001	6.2 ± 1.2	NA	NA
Systolic BP, mm Hg	122.9 ± 17.6	129.1 ± 17.4	<0.001	123.8 ± 17.7	126.3 ± 19.5	<0.001
Diastolic BP, mm Hg	78.1 ± 10.5	77.1 ± 10.4	<0.001	78.0 ± 10.5	75.8 ± 11.3	<0.001
Hypertension	39.0	63.8	<0.001	42.5	50.7	<0.001
Total cholesterol, mg/dL	194.2 ± 35.7	188.3 ± 41.1	<0.001	193.4 ± 36.6	NA	NA
Triglyceride, mg/dL	137.4 ± 101.3	177.5 ± 140.7	<0.001	142.9 ± 108.5	NA	NA
Hypercholesterolemia	15.2	26.9	<0.001	16.8	NA	NA
Hyperlipidemia	16.1	27.1	<0.001	17.6	NA	NA
Angina or MI	3.0	6.7	<0.001	3.5	5.7	<0.001
Previous stroke	2.4	5.8	<0.001	2.9	5.0	<0.001
Creatinine, mg/dL	0.82 ± 0.2	0.89 ± 0.4	<0.001	0.83 ± 0.2	NA	NA
Refractive error, D	−0.33 ± 1.9	−0.07 ± 1.7	<0.001	−0.29 ± 1.8	−0.01 ± 1.7	<0.001
Myopia, present vs. absent	34.1	30.4	<0.001	33.6	32.1	0.255
Hyperopia, present vs. absent	23.8	30.8	<0.001	24.8	31.8	<0.001
BMI, kg/m2	23.8 ± 3.0	24.9 ± 3.3	<0.001	23.9 ± 3.1	23.3 ± 3.5	<0.001
Obesity, present vs. absent	32.8	47.3	<0.001	34.8	31.5	0.013
Alcohol drinking	81.7	77.8	<0.001	81.2	73.9	<0.001
Current smoking	18.6	20.2	0.062	18.8	21.0	0.069
Ever smoking	40.3	48.2	<0.001	41.4	43.3	0.205
Vitamin D, ng/mL	19.3 ± 6.9	19.4 ± 7.1	0.809	19.3 ± 6.9	NA	NA

 Data are means ± SDs or frequency (%). ANOVA was used for continuous variables, and the χ² test was used for categoric variables. NA, not available.

A total of 1323 subjects (78.8%) reported that they had been diagnosed previously with DM (KDM), including 33 treated with only diet or exercise without medication, 1164 treated with oral hypoglycemic agents, 41 treated with insulin, and 85 treated with a combination of insulin and oral hypoglycemic agents, whereas 355 subjects (21.2%) were assigned to the NDM group. Type 1 DM occurred in only eight patients (0.4%) among those with DM > 40 years old. 

The prevalence of any DR in Koreans was 15.8% (95% CI, 14.1–17.5) in those 40+ years (Table 2). Of those aged 40+ years, VTDR was present in 4.6% (95% CI, 3.6–5.6) and CSME in 1.6% (95% CI, 1.0–2.2). The prevalence of any DR increased with age from those aged 40 to 49 years to those aged 60–69 years, and then decreased in those aged ≥70 years (P = 0.123). The prevalence of any DR was not different between males (15.4%) and females (16.2%) with DM (P = 0.639). 

Table 2

 

Prevalence and Severity of DR and ME According to Age and Sex in the Korean National Health and Nutrition Examination Survey, 2008 to 2011

	Sex, n (%, 95% CI)				Age Groups, n (%, 95% CI)
	Total, n = 1678	Male, n = 839	Female, n = 839	P	40–49 y, n = 215	50–59 y, n = 410	60–69 y, n = 588	70 + y, n = 465	P
Any DR	265 (15.8, 14.1–17.5)	129 (15.4, 13.0–17.8)	136 (16.2, 13.7–18.6)	0.639	22 (10.2, 6.2–14.2)	64 (15.6, 12.1–19.1)	108 (18.4, 15.3–21.5)	71 (15.3, 12.0–18.6)	0.123
DR grades
Minimal	114 (6.8, 5.5–8.0)	58 (6.9, 5.2–8.6)	56 (6.7, 5.0–8.4)	0.616	8 (3.7, 1.2–6.2)	26 (6.3, 3.9–8.7)	46 (7.8, 5.6–10.0)	34 (7.3, 4.9–9.7)	0.576
Mild	68 (4.1, 3.2–5.0)	29 (3.5, 2.3–4.7)	39 (4.6, 3.2–6.0)		7 (3.3, 0.9–5.7)	12 (2.9, 1.3–4.5)	33 (5.6, 3.7–7.5)	16 (3.4, 1.8–5.0)
Moderate	23 (1.4, 0.8–1.9)	14 (1.7, 0.8–2.5)	9 (1.1, 0.4–1.8)		2 (0.9, 0–2.2)	7 (1.7, 0.4–3.0)	8 (1.4, 0.5–2.3)	6 (1.3, 0.3–2.3)
Severe	14 (0.8, 0.4–1.2)	6 (0.7, 0.1–1.2)	8 (1.0, 0.3–1.7)		1 (0.5, 0–1.4)	4 (1.0, 0.0–2.0)	4 (0.7, 0.0–1.4)	5 (1.1, 0.2–2.0)
Proliferative	46 (2.7, 1.9–3.4)	22 (2.6, 1.5–3.6)	24 (2.9, 1.7–4.0)		4 (1.9, 0.1–3.7)	15 (3.7, 1.9–5.5)	17 (2.9, 1.5–4.3)	10 (2.2, 0.9–3.5)
ME	48 (2.8, 1.2–3.8)	26 (3.1, 1.9–4.3)	22 (2.6, 1.5–3.7)	0.660	6 (2.8, 0.6–5.0)	14 (3.4, 1.6–5.2)	12 (2.0, 0.9–3.1)	16 (3.4, 1.8–5.0)	0.915
CSME	27 (1.6, 1.0–2.2)	15 (1.8, 0.9–2.7)	12 (1.4, 0.6–2.1)	0.561	4 (1.9, 0.1–3.7)	8 (2.0, 0.6–3.4)	6 (1.0, 0.2–1.8)	9 (1.9, 0.7–3.1)	0.850
VTDR	77 (4.6, 3.6–5.6)	37 (4.4, 3.0–5.7)	40 (4.8, 3.4–6.2)	0.726	7 (3.3, 0.9–5.7)	25 (6.1, 3.8–8.4)	25 (4.3, 2.7–5.9)	20 (4.3, 2.5–6.1)	0.833

The presence and severity of DR were associated strongly with DM duration (see Figure). The prevalence of any DR in the NDM group was 2.8%, whereas the prevalence in those with a diabetes history ≥ 10 years was 33.2% (P < 0.001). Similarly, the prevalence of VTDR was significantly higher in those with DM for ≥10 years (12.5%) compared to NDM subjects (0.3%, P < 0.001). 

Figure

 

Prevalence of DR according to duration of DM in the Korean National Health and Nutrition Examination Survey, 2008 to 2011. PDR, proliferative diabetic retinopathy.

The prevalence of any DR was higher in the KDM group (19.3%) than in the NDM group (2.8%, P < 0.001, Table 3). Subjects with KDM had a higher frequency of all grades of retinopathy than did those with NDM (P < 0.001). The prevalence of VTDR was 5.8% in KDM and 0.3% in NDM (P < 0.001). 

Table 3

 

Prevalence of DR in Those With Known DM and NDM in the Korean National Health and Nutrition Examination Survey, 2008 to 2011

	n (%, 95% CI)
	Total, n = 1678 (%)	NDM, n = 355, 21.2%	KDM, n = 1323, 78.8%	P
Any DR	265 (15.8)	10 (2.8, 1.1–4.5)	255 (19.3, 17.2–21.4)	<0.001
DR grades
Minimal DR	114 (6.8)	6 (1.7, 0.4–3.0)	108 (8.2, 6.7–9.7)	<0.001
Mild DR	68 (4.1)	2 (0.6, 0–1.4)	66 (5.0, 3.8–6.2)
Moderate DR	23 (1.4)	1 (0.3, 0–2.6)	22 (1.7, 1.0–2.4)
Severe NPDR	14 (0.8)	1 (0.3, 0–0.9)	13 (1.0, 0.5–1.5)
PDR	46 (2.8)	0 (0)	46 (3.5, 2.5–4.5)
ME	48 (2.8)	6 (1.7, 0.4–3.0)	42 (3.2, 2.3–4.1)	0.136
CSME	27 (1.4)	0 (0)	27 (2.1, 1.3–2.9)	0.007
VTDR	77 (4.6)	1 (0.3, 0–0.9)	76 (5.8, 4.5–7.1)	<0.001

The association between various risk factors and either any DR or VTDR is shown in Table 4. In a multivariate analysis, independent risk factors for any DR were longer DM duration (OR, 1.08 per year increase), higher HbA1c (OR, 1.52 per 1% increase), higher systolic BP (OR, 1.02 per 1 mm Hg increase), higher creatinine (OR, 2.65 per 1 mg/dL increase), and lower BMI (OR, 0.91 per 1 unit). Similarly, VTDR was associated with longer DM duration (OR, 1.09 per year increase), higher HbA1c (OR, 1.64 per 1% increase), higher systolic BP (OR, 1.03 per 1 mm Hg increase), higher creatinine (OR, 2.33 per 1 mg/dL increase), and lower BMI (OR, 0.88 per 1 unit). 

Table 4

 

Risk Factors for DR and VTDR in the Korean National Health and Nutrition Examination Survey, 2008 to 2011

Characteristics	Any Diabetic Retinopathy				Vision-Threatening Diabetic Retinopathy
	Age–Sex Adjusted OR	P	Multivariate OR	P	Age–Sex Adjusted OR	P	Multivariate OR	P
Age, per y	1.01 (0.99–1.02)	0.148	0.97 (0.95–0.99)	0.021	1.00 (0.97–1.02)	0.913	0.93 (0.90–0.97)	0.003
Sex, female	1.04 (0.79–1.35)	0.797	1.21 (0.67–2.17)	0.533	1.09 (0.97–1.02)	0.872	3.31 (1.07–10.18)	0.037
Diabetes duration, y	1.10 (1.08–1.12)	<0.001	1.08 (1.06–1.10)	<0.001	1.13 (1.10–1.16)	<0.001	1.09 (1.05–1.13)	<0.001
Fasting glucose, mg/dL	1.01 (1.01–1.01)	<0.001	0.99 (0.99–1.00)	0.854	1.01 (1.01–1.01)	<0.001	0.99 (0.98–1.01)	0.688
HbA1c, %	1.57 (1.44–1.71)	<0.001	1.52 (1.28–1.80)	<0.001	1.57 (1.39–1.77)	<0.001	1.64 (1.24–2.18)	<0.001
Systolic BP, mm Hg	1.00 (0.99–1.01)	0.591	1.02 (1.01–1.03)	0.002	1.00 (0.98–1.01)	0.529	1.03 (1.00–1.05)	0.027
Diastolic BP, mm Hg	0.96 (0.96–0.98)	<0.001	0.97 (0.94–0.99)	0.002	0.95 (0.93–0.97)	<0.001	0.96 (0.92–1.01)	0.97
Hypertension*	0.79 (0.60–1.04)	0.090	0.88 (0.58–1.33)	0.546	0.64 (0.40–1.03)	0.067	0.59 (0.27–1.32)	0.206
Total cholesterol, mg/dL	1.00 (0.99–1.01)	0.392	1.00 (0.99–1.01)	0.603	0.99 (0.99–1.00)	0.274	0.99 (0.98–1.00)	0.114
Triglyceride, mg/dL	1.00 (1.00–1.00)	0.096	1.00 (1.00–1.00)	0.073	1.00 (0.99–1.00)	0.424	1.00 (1.00–1.00)	0.101
Hypercholesterolemia*	1.24 (0.92–1.66)	0.157	1.13 (0.76–1.66)	0.543	1.29 (0.78–2.13)	0.307	1.36 (0.68–2.72)	0.379
Angina or MI*	1.16 (0.71–1.91)	0.543	1.07 (0.55–2.10)	0.830	1.15 (0.48–2.73)	0.743	1.06 (0.31–3.74)	0.918
Previous stroke*	1.31 (0.76–2.25)	0.330	0.88 (0.43–1.84)	0.744	1.57 (0.65–3.78)	0.309	1.39 (0.37–5.12)	0.616
Creatinine, mg/dL	2.53 (1.512–4.26)	<0.001	2.65 (1.34–5.25)	0.005	3.34 (1.63–6.85)	<0.001	2.33 (1.03–5.31)	0.043
Myopia*	0.97 (0.72–1.32)	0.882	1.15 (0.70–1.88)	0.575	1.79 (1.09–2.96)	0.022	1.81 (0.76–4.32)	0.178
Hyperopia*	0.91 (0.67–1.24)	0.538	0.74 (0.46–1.19)	0.214	0.75 (0.41–1.35)	0.343	0.62 (0.24–1.58)	0.325
BMI, kg/m2	0.90 (0.86–0.94)	<0.001	0.91 (0.87–0.96)	0.001	0.84 (0.78–0.90)	<0.001	0.88 (0.80–0.96)	0.009
Alcohol drinking*	0.96 (0.68–1.36)	0.824	1.24 (0.79–1.93)	0.342	0.71 (0.40–1.26)	0.250	0.70 (0.32–1.52)	0.373
Current smoker*	1.23 (0.86–1.75)	0.251	1.18 (0.71–1.97)	0.509	1.44 (0.79–2.62)	0.227	1.34 (0.53–3.39)	0.530
Ever smoker*	1.07 (0.71–1.61)	0.751	0.82 (0.45–1.51)	0.535	1.21 (0.59–2.48)	0.602	1.80 (0.56–5.75)	0.319
Vitamin D, ng/mL	0.99 (0.97–1.01)	0.314	1.00 (0.98–1.03)	0.725	1.00 (0.96–1.03)	0.748	1.03 (0.98–1.07)	0.206

 Data are expressed as odds ratio and 95% confidence intervals.

*  Present versus absent.

Discussion

We have provided essential data on the prevalence and risk factors for DR in a representative Korean population. The prevalences of any DR and VTDR were 15.8% and 4.6%, respectively. When compared to population data in other Asian countries, the prevalence of DR in Koreans was lower than that of Chinese in Beijing (27.9%, ≥45 years), ¹⁶ Chinese in a rural area of Handan (43.1%, ≥30 years), ¹⁷ Indians in Singapore, (30.4%, ≥40 years), ¹⁸ and Malays in Singapore (35.0%, ≥40 years). ¹⁹ However, it was similar to that in urban India (17.6%, ≥20 years), ²⁰ and in another study in urban India (18.0%, ≥40 years). ²¹  

Our findings also were lower than those of a US population and Caucasians in Australia. They also were lower than that in Hispanics in the United States (46.9%, ≥40 years), ²² in National Health and Nutritional Examination Surveys (NHNES, 28.5%, ≥40 years), ²³ and in the Multi-ethnic Study of Atherosclerosis (MESA, 33.2%, ≥45 years). ²⁴ After stratifying by ethnic group, our data remained lower than that of Caucasians (24.8%–26.4%), blacks (36.7%–38.8%), and Hispanics (37.4%–46.9%) in NHNES and MESA. ^23,24 Finally, our data also are lower than in Caucasians in Australia (15.3%–32.8%). ^25–27  

Differences between epidemiologic studies may be attributable to methodologic differences, including differences in examination techniques and the grading system, or to differences in the characteristics of the study population. However, it was unlikely that seven standard photographs after pharmacologic pupil dilatation in our study, rather than two or four fundus photographs in other studies, ^{16–21,23–26} was the cause of the underestimation of the DR prevalence. In addition, the MESA consisted of a selectively healthy population free of clinical cardiovascular diseases, which may have resulted in underestimation of the prevalence of DR. 

The reason for this lower prevalence of DR in Korea is not known. However, one possible explanation is the high rate of DM diagnosis (78.8%). This contrasts with the results of other studies with the high prevalence of DR, of which the diagnostic rate in Beijing, China (65.9%) ¹⁶ and rural areas of China (36.0%) is low. ¹⁷ It is supported further by the fact that studies in India, with similarly low rates of DR as our data, reported a similarly high rate of DM diagnosis (75.2%–79.6%). ^20,21 These findings suggest that the screening and treatment for DM and DR in Korea are effective. A good diagnostic and education system for DM can reduce the incidence of DR through strict plasma glucose and BP control, which are well-known risk factors for DR. 

The DM duration, HbA1c, and systolic BP were confirmed to be independent risk factors in our study. This is consistent with previous reports. ^11,18,19,23 Moreover, these factors also were associated with VTDR. Our results supported the importance of these major risk factors in the Korean population. 

Less consistent risk factors, including lipid profile or obesity (BMI), were not associated with DR or VTDR in our study. In particular, BMI showed an inverse relationship with DR and VTDR in the multivariate regression analysis. This finding contrasted with many previous reports, in which obesity was a risk factor for DR. ^28–35 This may be due to the unique characteristics of type 2 DM in the Korean population; that is, many patients are nonobese and lose weight significantly during the course of the disease. 

Myopia or hyperopia was not associated with the DR in our study. This is in contrast with previous reports that patients with myopic refraction and a greater axial length are less likely to have DR, because axial elongation leads to decreased blood flow and reduced metabolic demand; thus, decreasing the impact of DM-induced microvascular changes. ^36–39 One possible reason for this result is that most previous studies were clinic-based, in contrast with our population-based study. However, an identical relationship was reported recently using population-based data from the Singapore Malay Eye Study. ³⁸ Thus, we cannot rule out the possibility that the pathology of DR in Korea differs and is not influenced by refractive status. In addition, because of the fact that myopia and hyperopia are defined simply as ±1 D in our study may be a limitation, further study with more comprehensive factors, such as high myopia or axial length is needed. 

The strength of this study was use of a nationwide, stratified, multistage, clustered sampling method, and standardized assessment method, which used seven standard photographs. Another is the large number of participants than that of previous population-based studies. ^16–27 One limitation was that we were unable to determine the type-specific prevalence and risk factors for DR due to the dearth of subjects with type 1 DM in Korea. Another limitation is that, considering that nonparticipants are older and, thus, more likely to have diabetes, this might result in underestimation. However, this effect may not be significant, given that the participation rate is over 90%. Finally, the cross-sectional design of the study prevented inference of causality. 

In summary, we report here important population-based data on the prevalence and risk factors for DR in a representative Korean population. The prevalences of any DR and VTDR in the Korean population were 15.8% and 4.6%, respectively; these are lower than those of other Asian and Western populations. The conditions of DR and VTDR were associated with a longer DM duration, poor glycemic control, and hypertension. 

Acknowledgments

Supported by Research Grants from Industry Academic Cooperation Foundation at Catholic University of Korea, 2013 (5-2013-D0383-00001). The authors alone are responsible for the content and writing of the paper. The English in this document has been checked by at least two professional editors, both native speakers of English. For a certificate, please see: http://www.textcheck.com/certificate/4jl6uK. 

Disclosure: D. Jee, None; W.K. Lee, None; S. Kang, None