Diabetic Retinopathy

Prevalence and Risk Factors forDiabetic Retinopathy

The Singapore Malay Eye Study

Tien Y. Wong, MD, PhD,1,2,3 Ning Cheung, MBBS,1 Wan Ting Tay, BSc,2 Jie Jin Wang, PhD,1,4Tin Aung, MD, PhD,2,3 Seang Mei Saw, PhD,2,3,5 Su Chi Lim, MD,6 E. Shyong Tai, MD,7Paul Mitchell, MD, PhD4

Purpose:Design:

To describe the prevalence and risk factors of diabetic retinopathy in Asian Malays.Population-based cross-sectional study.

Participants:

Persons with diabetes of Malay ethnicity aged 40 to 80 years in Singapore.

Methods:

Diabetes mellitus was dened as random glucose of 11.1 mmol/l or more, use of diabetic

medication, or a physician diagnosis of diabetes. Retinal photographs taken from both eyes were graded fordiabetic retinopathy using the modied Airlie House classication system.

Main Outcome Measures:retinopathy.

Any diabetic retinopathy, retinopathy grades, macular edema, or vision-threatening

Results:

Of the 3261 persons who participated in this study, 757 (23.1%) had diabetes and gradable retinal

photographs. In persons with diabetes, the overall prevalence of any retinopathy was 35.0% (95% condenceinterval [CI], 28.2 43.4), the overall prevalence of macular edema was 5.7% (95% CI, 3.29.9), and the overallprevalence of vision-threatening retinopathy was 9.0% (95% CI, 5.8 13.8). Compared with men, women hadsignicantly higher proportions with more severe diabetic retinopathy, moderate (12% vs. 4%) and severe (3%vs. 0.2%) nonproliferative retinopathy, proliferative retinopathy (7% vs. 2%), and vision-threatening retinopathy(13% vs. 3%). In multiple logistic regression, independent risk factors for any retinopathy were longer diabetesduration (odds ratio [OR], 1.07; 95% CI, 1.04 1.09, per year increase), higher hemoglobin A1C (OR, 1.21; 95% CI,1.10 1.33, per % increase), hypertension (OR, 1.85; 95% CI, 1.04 3.30), and higher pulse pressure (OR, 1.34,95% CI, 1.19 1.51, per 10-mmHg increase); older age (OR, 0.73; 95% CI, 0.57 0.93, per decade increase) andhigher total cholesterol levels (OR, 0.75; 95% CI, 0.63 0.89, per 1-mmol/l increase) were protective of anyretinopathy. Vision-threatening retinopathy additionally was associated with previous stroke (OR, 3.74; 95% CI,1.24 11.26), cardiovascular disease (OR, 2.23; 95% CI, 1.08 4.62), and chronic kidney disease (OR, 4.45; 95%CI, 2.18 9.07). Female gender was not an independent risk factor for diabetic retinopathy after adjusting formetabolic and socioeconomic risk factors.

Conclusions:

One in 10 Malay adults with diabetes in Singapore has vision-threatening diabetic retinopathy.

Risk factors for retinopathy in this population are largely similar to white populations elsewhere, suggesting thatcontrol of these risk factors may reduce both the prevalence and impact of retinopathy.

Financial Disclosure(s):

The authors have no proprietary or commercial interest in any materials discussed

in this article. Ophthalmology 2008;115:1869 1875 2008 by the American Academy of Ophthalmology.

As a leading cause of blindness in working-aged people,diabetic retinopathy is a serious threat to quality of life formillions worldwide.1,2 However, current estimates of theprevalence and risk factors for diabetic retinopathy aremostly derived from studies of non-Asian populations.320To date, population-based data on diabetic retinopathy inAsians remain scarce.21Only 2 population-based studies examined diabetic reti-nopathy using standardized protocols based on assessmentof retinal photographs in Asians. The Chennai Urban RuralEpidemiological Study (CURES) reported a retinopathy prev-alence rate of 17.6% in India.22 More recently, the Multi-Ethnic Study of Atherosclerosis (MESA) showed that 25.7%of Chinese Americans in the United States had diabetic reti-

nopathy.23 These prevalence rates were notably lower thanthose found in studies of white persons. It has been hypothe-sized that these ethnic differences may result from varyingsusceptibilities to diabetic microvascular complications.24Apart from visual morbidity, there is now good evidenceto suggest that the presence of diabetic retinopathy alsosignies an increased risk of systemic vascular complica-tions, including stroke, coronary heart disease, heart failure,and nephropathy.24 28 These data indicate that diabetic ret-inopathy reects widespread microcirculatory dysfunctionnot only in the eye, but also in vital organs elsewhere in thebody.24 Thus, a better understanding of the systemic riskfactors of diabetic retinopathy in Asians may uncover im-portant insights into the ethnic variability in the risk for

2008 by the American Academy of OphthalmologyPublished by Elsevier Inc.

ISSN 0161-6420/08/$see front matterdoi:10.1016/j.ophtha.2008.05.014

1869

Ophthalmology Volume 115, Number 11, November 2008

systemic complications of diabetes. This study examinedthe prevalence and risk factors of diabetic retinopathy in apopulation-based sample of middle-aged and older AsianMalays in Singapore.

Patients and MethodsStudy PopulationThe Singapore Malay Eye Study (SiMES) is a population-based, cross-sectional study of 3280 Malay adults, aged be-tween 40 and 79 years. Details of the SiMES design, samplingplan, and methodology have been reported elsewhere.29 34 Inbrief, an age-stratied random sampling of all Malay adultsresiding in the southwestern part of Singapore was performed,where 1400 names from each decade (40 49, 50 59, 60 69,and 70 79 years), or an initial 5600 names, were selected. Ofthese, 4168 individuals (74%) were determined to be eligible toparticipate. A potential participant was considered to be ineli-gible if the person had moved from the residential address, hadnot lived there in the past 6 months, was deceased, or wasterminally ill. Of 4168 eligible individuals, 3280 participantstook part in the study (78.7% participation rate). Of the non-participants, 831 (20% of eligible participants) declined toparticipate and 57 (1%) were not contactable. Nonparticipantstended to be in the older age group (70 79 years of age) incomparison with participants, but there were few differences ingender, sampling location, and telephone ownership betweenthe 2 groups (data not shown). Of these 3280 participants, thosewho did not undergo retinal photography or those with retinalphotographs of insufcient quality for grading were excluded (n19), leaving 3261 participants for analyses. In the primaryanalyses, the study focused on diabetic participants (n 757)only. Diabetes mellitus was dened as random glucose of 11.1mmol/l or more, use of diabetic medication, or a physiciandiagnosis of diabetes.30 The hospital institutional review boardapproved the study, conducted in accordance with the Declara-tion of Helsinki, with written informed consent obtained fromall participants.

Retinal Photography and Diabetic RetinopathyAssessmentRetinal photography was performed following a standardized proto-col.23,29 After pupil dilation, 2 retinal photographs, centered at theoptic disc and macula, were obtained from both eyes of each partic-ipant using a digital retinal camera (Canon CR-DGi with a 10-D SLRback; Canon, Tokyo, Japan). Photographs then were sent to theUniversity of Sydney and were graded for retinopathy and otherretinal diseases based on the Blue Mountains Eye Study protocol.18,35Retinopathy was considered present if any characteristic lesion asdened by the Early Treatment Diabetic Retinopathy Study severityscale was present: microaneurysms (MA), hemorrhages, cotton woolspots, intraretinal microvascular abnormalities (IRMA), hard exudates(HE), venous beading, and new vessels.23 For each eye, a retinopathyseverity score was assigned according to a scale modied from theAirlie House classication system: level 10, no retinopathy present;level 15, hemorrhage present without any denite MA; level 20, MAonly with no other retinopathy lesions present; level 31, MA and 1 ormore of hemorrhage or MA less than standard photograph 2A, hardexudates, venous loops, questionable cotton wool spots, IRMA, orvenous beading; level 41, MA and 1 or more of cotton wool spotsor IRMA less than standard photograph 8A; level 51, MA and 1 ormore of venous beading, hemorrhage, MA of 2A or more, or

1870

IRMA of 8A or more; level 60, brous proliferation with no otherproliferative lesions; level 61 through 64, laser scatter photocoag-ulation scars with retinopathy levels 31 through 51; level 65,proliferative diabetic retinopathy less than high-risk characteris-tics, as dened in the Diabetic Retinopathy Study; level 70, pro-liferative diabetic retinopathy with high-risk characteristics; andlevel 80, total vitreous hemorrhage.Macular edema was dened by hard exudates in the presence ofMA and blot hemorrhage within 1 disc diameter from the fovealcenter or presence of focal photocoagulation scars in the maculararea. Clinically signicant macular edema (CSME) was consideredpresent when the macular edema involved was within 500 m ofthe foveal center or if focal photocoagulation scars were present inthe macular area.Four primary outcomes for this study were dened on the basisof the severity scores of the worse of the 2 eyes. Retinopathyseverity was categorized as minimal nonproliferative diabetic ret-inopathy (NPDR; level 15 through 20), mild NPDR (level 35),moderate NPDR (level 43 through 47), severe NPDR (level 53),and proliferative retinopathy (level more than 60). Macular edemawas dened as present and absent and was divided further aspresence or absence of CSME. Vision-threatening retinopathy wasdened as the presence of severe NPDR, proliferative retinopathy,or CSME using the Eye Diseases Prevalence Research Groupdenition.36 If an eye was ungradable, the scores for the fellow eyewere used to dene these outcomes.

Assessment and Denitions of Risk FactorsAll participants underwent a standardized interview and examina-tion, including collection of blood samples.29,30,34 Socioeconomicmeasures (e.g., income, education), lifestyle risk factors (e.g.,smoking), medication use, and self-reported history of systemicdiseases were elicited from the interview. Previous myocardialinfarction and stroke were ascertained from self-report. Cardiovas-cular disease was dened as history of previous myocardial infarc-tion, angina, or stroke. Overweight was dened as body massindex (BMI) of more than 25 kg/m2. Blood pressure was measuredaccording to the protocol used in the Multi-Ethnic Study of Ath-erosclerosis.23 Hypertension was dened as systolic blood pressure140 mmHg or more, diastolic blood pressure of 90 mmHg or more,or use of antihypertensive medication. Nonfasting venous bloodsamples were drawn and sent for analysis of serum lipid levels(total cholesterol, high-density lipoprotein cholesterol, and low-density lipoprotein cholesterol), hemoglobin A1C, creatinine, andglucose at the National University Hospital Reference Laboratoryon the same day. Hyperlipidemia was dened as total cholesterolof 6.2 mmol/l or more or use of lipid-lowering drugs. Urinesamples were collected to determine levels of microalbuminuriaand creatinine at the Alexandra Hospital Laboratory. Chronickidney disease was dened as estimated glomerular ltration rateof less than 60 ml/minute per 1.73 m2.34

Statistical AnalysisAny diabetic retinopathy and vision-threatening retinopathy wereanalyzed as binary outcome variables. All potential risk factorswere analyzed either as binary traits (e.g., hypertension) or lineartraits for continuous variables (e.g., BMI). To test threshold ef-fects, the continuous variables were analyzed without a clinicallydistinct cutpoint as more than versus less than the 90th percentile(e.g., vibration perception threshold). Logistic regression modelswere used to estimate the odds ratios (ORs) and 95% condenceintervals (CIs) for both diabetic retinopathy and vision-threateningretinopathy for each risk factor, adjusting for age and gender. Inmultivariate analysis, the authors further adjusted for socioeco-

Wong et al

Diabetic Retinopathy in an Asian Population

nomic and metabolic risk factors. To evaluate the extent thatmetabolic and socioeconomic risk factors may explain the excessprevalence of retinopathy in men and women, the percent reduc-tion in odds associated with adjustment for these factors wasestimated according to the following formula: (ra rb)/(ra 1) 100, where ra is the OR of retinopathy in men comparedwith whites adjusted for age only (model 1) and rb is the OR afterfurther adjustment in model 2 (metabolic risk factors) and model 3

longer diabetes duration (OR, 1.07 per year increase), higherhemoglobin A1C (OR, 1.21 per 1% increase), hypertension (OR,1.85), and higher pulse pressure (OR, 1.34 per 1-mmHg increase).Older age (OR, 0.73 per 10-year increase) and higher total cho-lesterol levels (OR, 0.75 per 1-mmol/l increase) were associatedwith reduced odds of any retinopathy. Vision-threatening retinop-athy additionally was associated with self-reported stroke (OR,3.74) and chronic kidney disease (OR, 4.45). The prevalence of

(socioeconomic risk factors). All analyses were performed using

chronic kidney disease in this study was 34.5% (n

262), of

SPSS software version 15 (SPSS, Inc., Chicago, IL).

which 42.5% (n

110) had diabetic retinopathy. Female gender was

not associated independently with diabetic retinopathy in multivariateanalysis. Although myocardial infarction was associated with higher

Results

Table 1 shows the characteristics of the study population. Com-pared with nondiabetic subjects, diabetic participants were morelikely to be older, to be female, and to have higher blood pressure,BMI, and serum glucose, but were less likely to be smokers.Among those with diabetes, women were more likely to beyounger, to be nonsmokers, and to have lower education levels,higher BMI, and higher levels of lipids than men. In this popula-tion, 92 persons (12.2%) had undiagnosed diabetes. Among par-ticipants with diabetes, 83 (11.0%) reported previous myocardialinfarction, 45 (6.0%) reported having a previous stroke, and 141(18.7%) had prevalent cardiovascular disease.Table 2 shows that the age-standardized overall prevalence ofany diabetic retinopathy in the population was 35.0% (95% CI,28.2 43.4). Vision-threatening retinopathy was present in 9.0%(95% CI, 5.8 13.8), macular edema was present in 5.7%, andCSME was present in 3.5% of participants. Compared with men,women had signicantly higher proportions, with moderate(12.3% vs. 3.8%) and severe (3.3% vs. 0.2%) NPDR, proliferativeretinopathy (6.9% vs. 2.0%), and vision-threatening retinopathy(12.9% vs. 3.0%).Table 3 shows the associations of various risk factors witheither any or vision-threatening diabetic retinopathy. In multivar-iate analysis, independent risk factors for any retinopathy were

odds of retinopathy, especially for vision-threatening retinopathy(OR, 2.29), this association was not statistically signicant (P0.08).Table 4 shows increased age-adjusted odds of diabetic retinop-athy in women compared with men. This association attenuatedand was no longer statistically signicant after further adjustmentsfor metabolic and socioeconomic risk factors. The reduction inodds of retinopathy associated with gender, with adjustment ofspecic risk factors (model 1 to model 3), also was evaluated.After adjusting for metabolic risk factors, the ORs of retinopathyassociated with gender, age-adjusted OR of 1.3 for any retinopa-thy, OR of 2.1 for macular edema, and OR of 1.9 for vision-threatening retinopathy, were reduced by 33%, 45%, and 22%,respectively. Further adjustment for socioeconomic factors re-duced the excess prevalence of any retinopathy even more (100%),but the prevalence of macular edema and vision-threatening reti-nopathy remained the same.

Discussion

Knowledge regarding the epidemiology of diabetic retinop-athy in Asians remains limited because of the paucity ofpopulation-based data.21 The current study provides newdata on the prevalence and risk factors for diabetic retinop-

Table 1. Participant Characteristics in the Singapore Malay Eye Study

Persons without

Persons with

Men with

Women with

All Persons

Diabetes

Diabetes

Diabetes

Diabetes

Characteristic

(N

3261)

(n

2504)

(n

757)

P Value*

(n

329)

(n

428)

P Value

Age (yrs)Male genderIncome ( S$1000)Education (elementary or less)HbA1C (%)Systolic BP (mmHg)HypertensionTotal cholesterol (mmol/l)HDL cholesterol (mmol/l)Triglycerides (mmol/l)HyperlipidemiaPrevious myocardial infarctionPrevious strokeCardiovascular diseaseBody mass index (kg/m2)Current smoker

58.7 (11.01)1570 (51.9)1816 (55.4)2444 (75.2)6.5 (1.56)147.1 (23.76)2233 (68.4)5.6 (1.17)1.4 (0.33)1.6 (1.33)1303 (40.6)210 (6.5)81 (2.5)363 (11.2)26.4 (5.12)659 (20.2)

57.5 (11.19)1240 (49.5)1341 (53.6)1797 (71.9)5.9 (0.60)144.8 (23.41)1583 (63.2)5.7 (1.13)1.4 (0.34)1.5 (1.23)907 (36.8)127 (5.1)36 (1.4)222 (8.9)26.0 (5.16)562 (22.5)

62.5 (9.42)330 (43.4)475 (62.7)647 (85.8)8.4 (2.03)154.5 (23.39)650 (85.5)5.5 (1.29)1.3 (0.31)1.9 (1.58)396 (52.9)83 (11.0)45 (6.0)141 (18.7)27.5 (4.81)97 (12.8)

0.0010.0030.0010.0010.0010.0010.0010.0010.0010.0010.0010.0010.0010.0010.0010.001

63.6 (9.88)165 (50.2)259 (79.0)8.3 (1.95)151.2 (21.39)280 (84.8)5.2 (1.17)1.1 (0.25)1.8 (1.48)143 (44.3)49 (14.9)17 (5.2)67 (20.4)26.4 (4.36)88 (26.7)

61.7 (8.97)310 (72.4)388 (91.1)8.5 (2.09)157.1 (24.52)370 (86.0)5.7 (1.33)1.4 (0.31)1.9 (1.65)253 (59.5)34 (8.0)28 (6.6)74 (17.4)28.2 (4.99)9 (2.1)

0.0060.0010.0010.190.0010.640.0010.0010.490.0010.0030.420.290.0010.001

BP

blood pressure; HbA1C

hemoglobin A1C; HDL

high-density lipoprotein.

Data presented are means (standard deviations) or number (%), as appropriate for variable.*P value for difference in characteristics by diabetes status, based on chi-square test or t test, as appropriate.

P value for the difference in characteristics by gender, based on chi-square test or t test, as appropriate.

1871


Table 2. Prevalence and Severity of Diabetic Retinopathy and Macular Edema, by Gender and Diabetes Status

Newly

Persons withDiabetes

Men withDiabetes

Women withDiabetes

DiagnosedDiabetes

KnownDiabetes

(n

757),

(n

329),

(n

428),

(n

92),

(n

665),

% (95%Condence

% (95%Condence

% (95%Condence

% (95%Condence

% (95%Condence

Interval)

Interval)

Interval)

P Value*

Interval)

Interval)

P Value

Any retinopathy

35.0 (28.243.4)

22.9 (15.833.9)

42.8 (32.855.6)

0.05

28.6 (15.349.8)

35.3 (27.745.0)

0.003

Retinopathy grades

NoneMinimalMildModerateSevereProliferativeMacular edemaCSMEVision-threatening

65.0 (55.476.1)11.2 (7.616.3)7.9 (4.613.0)8.9 (5.713.8)2.1 (0.65.7)4.9 (2.78.8)5.7 (3.29.9)3.0 (1.76.1)9.0 (5.813.8)

77.1 (60.098.4)10.1 (5.519.1)6.7 (2.615.8)3.8 (2.310.1)0.2 (0.06.7)2.0 (1.18.2)1.6 (0.87.8)1.3 (0.67.6)3.0 (1.89.2)

57.2 (46.370.6)11.8 (7.119.3)8.6 (4.216.1)12.3 (7.220.2)3.3 (0.99.1)6.9 (3.413.3)8.4 (4.415.4)4.2 (2.09.3)12.9 (7.820.8)

0.05

0.030.090.03

71.4 (51.997.6)16.7 (7.134.4)0.6 (0.011.0)10.4 (3.026,7)0.00.8 (0.011.3)0.8 (0.011.3)0.8 (0.011.3)0.8 (0.011.3)

64.7 (53.777.7)9.2 (6.114.4)9.8 (5.516.6)7.8 (4.813.0)2.6 (0.77.4)5.8 (3.011.0)6.8 (3.612.5)3.4 (1.77.5)10.8 (6.717.2)

0.02

0.040.100.003

retinopathy

CSME

clinically signicant macular edema.

*P value for the difference in prevalence and severity of retinopathy by gender, based on chi-square test.

P value for the difference in prevalence and severity of retinopathy by diagnosis of diabetes, based on chi-square test.Age-standardized to 2000 Malay population census.

athy in Asian Malays, which are the third largest distinctracial or ethnic group in Asia, including 300 to 400 millionpeople. This study found an overall prevalence of diabeticretinopathy of 35%, but perhaps more importantly, approx-

imately 1 in 10 persons with diabetes had vision-threateningretinopathy. The presence of diabetic retinopathy was asso-ciated with longer diabetes duration, poorer glycemic andblood pressure control, and lower levels of total and LDL

Table 3. Risk Factors for Diabetic Retinopathy and Vision-Threatening Retinopathy

Any Retinopathy

Vision-Threatening Retinopathy

Age-GenderOdds Ratio*(95% Condence

MultivariableOdds Ratio(95% Condence

Age-GenderOdds Ratio*(95% Condence

Multivariate OddsRatio (95%Condence

CharacteristicsAge (per 10 yrs)Diabetes duration (per year)Serum glucose (per mmol/l)HbA1C (per mmol/l)Systolic BP (per 10 mmHg)Pulse pressure (per 10 mmHg)HypertensionTotal cholesterol (per mmol/l)HDL cholesterol (per mmol/l)LDL cholesterol (per mmol/l)Triglycerides (per mmol/l)HyperlipidemiaBody mass index (per kg/m2)Previous myocardial infarctionPrevious strokeCardiovascular diseaseChronic kidney diseaseInsulin useOral diabetic medicationCurrent smoker

Interval)0.93 (0.791.09)1.07 (1.051.09)1.05 (1.021.08)1.19 (1.111.29)1.15 (1.071.23)1.30 (1.181.43)1.64 (1.032.60)0.90 (0.791.01)0.87 (0.511.49)0.80 (0.690.94)1.06 (0.961.16)0.88 (0.651.19)0.96 (0.930.99)1.34 (0.832.15)0.81 (0.421.56)1.15 (0.791.68)1.76 (1.272.45)4.05 (1.938.53)1.21 (0.891.63)1.07 (0.651.76)

P Value0.380.0010.0010.0010.0010.0010.040.080.610.0050.260.390.020.230.540.470.0010.0010.230.79

Interval)0.73 (0.570.93)1.07 (1.041.09)1.05 (1.021.09)1.21 (1.101.33)1.17 (1.081.28)1.34 (1.191.51)1.85 (1.043.30)0.75 (0.630.89)1.18 (0.622.26)0.71 (0.580.87)0.93 (0.831.05)0.76 (0.521.10)0.96 (0.921.00)1.57 (0.882.81)1.06 (0.482.34)1.22 (0.771.94)1.48 (0.992.21)2.03 (0.864.78)1.03 (0.701.53)1.28 (0.692.36)

P Value0.010.0010.0040.0010.0010.0010.040.0010.610.0010.260.150.080.130.880.400.060.110.880.44

Interval)0.95 (0.731.23)1.08 (1.051.11)1.09 (1.051.14)1.18 (1.051.32)1.28 (1.141.42)1.59 (1.371.86)1.76 (0.774.01)1.26 (1.051.51)0.78 (0.321.90)1.12 (0.891.41)1.22 (1.081.37)2.01 (1.183.40)0.92 (0.870.98)1.93 (0.983.82)1.74 (0.744.08)1.94 (1.12,3.35)4.21 (2.467.22)1.83 (0.684.94)0.90 (0.551.46)0.46 (0.151.36)

P Value0.690.0010.0010.0060.0010.0010.180.010.590.350.0020.010.0060.060.200.020.0010.230.700.16

Interval)0.61 (0.400.94)1.08 (1.051.11)1.10 (1.051.17)1.23 (1.061.42)1.35 (1.181.55)1.73 (1.422.11)2.2 (0.796.26)1.12 (0.881.42)0.91 (0.332.49)1.07 (0.801.44)1.08 (0.921.27)1.83 (0.953.52)0.93 (0.861.00)2.29 (0.905.83)3.74 (1.2411.26)2.23 (1.084.62)4.45 (2.189.07)0.48 (0.131.74)0.69 (0.371.31)0.81 (0.242.74)

P Value0.030.0010.0010.0070.0010.0010.130.360.850.650.370.070.040.080.020.030.0010.270.260.81

BP

blood pressure; HbA1C

hemoglobin A1C; HDL

high-density lipoprotein; LDL

low-density lipoprotein.

*Adjusted for age and gender.

Adjusted for age, gender, metabolic risk factors (body mass index, HbA1C, duration of diabetes, and systolic blood pressure), and socioeconomic factors

(income, housing, and education).

1872

Wong et al


Table 4. Effect of Potential Explanatory Factors on the Excess Prevalence of Retinopathy in Women Compared with Men

Any RetinopathyOdds Ratio

Macular EdemaOdds Ratio

Vision-Threatening RetinopathyOdds Ratio

(95% CondenceInterval),

% ReductionExcess


% ReductionExcess


% ReductionExcess

Adjustment (Models)Age (1)Plus metabolic risk factors (2)Plus socioeconomic factors (3)

Women vs Men*1.3 (1.01.8)1.2 (0.81.7)1.0 (0.61.6)

PrevalenceReference33%100%

Women vs Men*2.1 (1.14.2)1.6 (0.73.5)1.6 (0.73.7)


Women vs Men*1.9 (1.13.2)1.7 (0.93.4)1.7 (0.83.5)


*Odds ratio (95% condence interval) of various retinopathy outcomes in association with gender adjusting for age (model 1), metabolic risk factors (bodymass index, hemoglobin A1c, duration of diabetes, and systolic blood pressure; model 2), and socioeconomic factors (income, housing, and education;model 3).

Percent reduction in excess prevalence dened by the formula: (ra rb)/(ra 1), where ra is the odds ratio of retinopathy in women versus men, adjusted

for age only (model 1, reference) and rb is the odds ratio after additional adjustment for the variables in models 2 and 3.

cholesterol. Vision-threatening retinopathy also was associatedwith systemic vascular diseases, including stroke and a secondmajor microvascular complication, chronic kidney disease.Although there have been other population-based studieson diabetic retinopathy in Asian populations,21 only 2 pre-vious studies used retinal photography to ascertain preciselythe presence of retinopathy.22,23 The CURES in Asian In-dian persons reported a prevalence of 17.5%, and theMESA, which was conducted in the United States, reporteda prevalence of 25.7% among Chinese American partici-pants with diabetes. The authors are unaware of any previ-ous population-based studies on the prevalence of diabeticretinopathy in other Asian populations. Although the cur-rent prevalence rates are notably higher than those reportedin the CURES and the MESA, methodologic differencesand differences in study population characteristics limitdirect comparison of results. For example, participants inthe CURES had lower blood pressure (mean systolic bloodpressure, 130 mmHg in CURES vs. 155 mmHg in SiMES)and a lower prevalence of hypertension (48.7 vs. 85.5%)and myocardial infarction (5.2% vs. 11.0%).22 The MESAalso consisted of a selectively healthy population free ofclinical cardiovascular diseases.23 These differences mayform the basis for the lower prevalence rates reported fromthe CURES and MESA as compared with the SiMES.The retinopathy prevalence rates reported from this studyare more comparable with existing population-based studiesin non-Asian populations, as shown in the recent metaanaly-sis of 7 population-based studies of whites, blacks, andHispanics. This metaanalysis found an overall retinopathyprevalence of 35.8% and vision-threatening retinopathyprevalence of 7.3%, similar to those found in this study (seeFig 1).36 The cross-sectional survey of diabetic retinopathyin the United Arab Emirates reported a lower prevalencerate of retinopathy (19.0%; Fig 1).20An observation that has important public health implica-tions in this study is the higher prevalence of more severediabetic retinopathy in women than in men. Previous studieshave not shown a consistent pattern of gender variation indiabetic retinopathy prevalence.7,37 However, the associa-tion of female gender and vision-threatening retinopathyseems to be related to gender differences in the distributionof retinopathy risk factors, because differences between

women and men became attenuated and lost their statisticalsignicance in analyses adjusted for metabolic and socioeco-nomic risk factors. This suggests that the higher prevalence ofsevere retinopathy in women may be reduced by greater atten-tion to the appropriate attention and management of modiablerisk factors in lower socioeconomic groups.Epidemiologic studies, largely among white persons,have identied the major risk factors for diabetic retinopa-thy.319 Of these, the most consistent risk factors are longerduration of diabetes, hyperglycemia (and poorer glycemiccontrol), and hypertension.7,18,19 The current data extendthese key observations to an Asian population. In addition,also consistent with previous studies of older people withtype 2 diabetes,38 41 an association between younger ageand diabetic retinopathy was observed.Associations of diabetic retinopathy with other less con-sistent risk factors, such as obesity and hyperlipidemia, alsohave been documented.4,5,9,10,13,14,42 Interestingly, associa-tions of lower BMI (with marginal statistical signicance)and lower total (and low-density lipoprotein) cholesterollevels with diabetic retinopathy also were found in thisstudy. Similarly, some previous population-based studies inwhites, including the Wisconsin Epidemiological Study ofDiabetic Retinopathy, also have demonstrated an inverserelationship between BMI and the prevalence or severity ofdiabetic retinopathy.7,42,43 The mechanisms underlying thisassociation are unclear but have been linked to the adverseeffects of malnutrition on the microvascular system.42 Thending of an association between lower cholesterol levelswith diabetic retinopathy, however, remains unexplainedand requires further exploration by other studies.Finally, this study documents associations of diabeticretinopathy with systemic vascular complications of diabe-tes, such as stroke and chronic kidney disease. These asso-ciations were conned mainly to those with more severeretinopathy (e.g., vision-threatening retinopathy). Thesendings are in keeping with the extant literature based ondata from non-Asian population-based studies.24 28,44,45Strengths of this study include its population-based sam-ple, masked grading of retinal photographs from both eyesat the University of Sydney using the Blue Mountains EyeStudy protocol, high frequency of gradable photographs,and detailed assessment of risk factors. Limitations also

1873


Of importance, the major risk factors for retinopathy in thispopulation largely are similar to those reported in whitepopulations. This suggests that aggressively addressingthese risk factors may reduce the public health impact andburden of this preventable cause of vision loss in Asia.

References

1. Mohamed Q, Gillies MC, Wong TY. Management of diabeticretinopathy: a systematic review. JAMA 2007;298:90216.2. Frank RN. Diabetic retinopathy. N Engl J Med 2004;350:48 58.3. UK Prospective Diabetes Study Group. Tight blood pressurecontrol and risk of macrovascular and microvascular complica-tions in type 2 diabetes: UKPDS 38. BMJ 1998;317:70313.4. Chew EY, Klein ML, Ferris FL III, et al. Association ofelevated serum lipid levels with retinal hard exudate in diabeticretinopathy. Early Treatment Diabetic Retinopathy Study(ETDRS) report 22. Arch Ophthalmol 1996;114:1079 84.5. Klein BE, Moss SE, Klein R, Surawicz TS. The WisconsinEpidemiologic Study of Diabetic Retinopathy. XIII. Relation-ship of serum cholesterol to retinopathy and hard exudate.Ophthalmology 1991;98:12615.

Figure 1. Bar graphs demonstrating the prevalence of diabetic retinopa-thy in the Singapore Malay Eye Study and other population-based studies.BDES Beaver Dam Eye Study; BMES Blue Mountains Eye Study;VIP Visual Impairment Project; VER Vision Evaluation Research;SAHS San Antonio Heart Study; SLVDS San Luis Valley DiabetesStudy; WESDR Wisconsin Epidemiologic Study of Diabetic Retinop-athy; SiMES Singapore Malay Eye Study; UAE Cross SectionalSurvey of Diabetic Retinopathy in the United Arab Emirates.

should be noted. First, the cross-sectional design of thestudy prevented inferring causality. Second, retinopathy le-sions in this study were ascertained from 2 digital imagesper eye, a simpler methodology than that used in some ofthe previous population-based studies.7,17,18,22 Thus, thepossibility of underestimating prevalence cannot be totallyexcluded. However, such underestimation, if any, is un-likely to be substantial because the prevalence rates in factare similar or higher than those of other studies that usedretinal photographs of more than 2 elds to ascertain reti-nopathy.22,36 Third, diabetes was dened based on randomblood glucose levels (and other criteria) rather than fastingglucose or oral glucose tolerance test results, which aremore widely recognized methods of diagnosis for diabetes.Therefore, there could be some misclassications on diabe-tes status. Fourth, because this study used nonstereoscopicdigital photography, the denition of CSME may not becomparable directly with studies that used a different pro-tocol (e.g., stereoscopic photography used in the EarlyTreatment Diabetic Retinopathy Study). Finally, given thelarge number of risk factors examined, some unexpectedndings may be the result of chance (e.g., association ofretinopathy with lower cholesterol levels).In summary, this study provides key population-baseddata on the epidemiologic characteristics of diabetic reti-nopathy in Asians. The data show that approximately 1 in 3persons of Asian Malay ethnicity had diabetic retinopathyand that 1 in 10 had vision-threatening diabetic retinopathy.

1874

6. Klein R, Klein BE, Moss SE, Cruickshanks KJ. Relationshipof hyperglycemia to the long-term incidence and progressionof diabetic retinopathy. Arch Intern Med 1994;154:2169 78.7. Klein R, Klein BE, Moss SE, et al. The Wisconsin Epidemi-ologic Study of Diabetic Retinopathy. III. Prevalence and riskof diabetic retinopathy when age at diagnosis is 30 or moreyears. Arch Ophthalmol 1984;102:52732.8. Klein R, Klein BE, Moss SE, et al. Glycosylated hemoglobin pre-dicts the incidence and progression of diabetic retinopathy. JAMA1988;260:2864 71.9. van Leiden HA, Dekker JM, Moll AC, et al. Risk factors forincident retinopathy in a diabetic and nondiabetic population:the Hoorn study. Arch Ophthalmol 2003;121:24551.10. Ferris FL III, Chew EY, Hoogwerf BJ, Early Treatment Dia-betic Retinopathy Study Research Group. Serum lipids anddiabetic retinopathy. Diabetes Care 1996;19:12913.11. Haffner SM, Fong D, Stern MP, et al. Diabetic retinopathy inMexican Americans and non-Hispanic whites. Diabetes 1988;37:878 84.12. Hamman RF, Mayer EJ, Moo-Young GA, et al. Prevalenceand risk factors of diabetic retinopathy in non-Hispanic whitesand Hispanics with NIDDM. San Luis Valley Diabetes Study.Diabetes 1989;38:12317.13. Klein R, Klein BE, Moss SE. Is obesity related to microvas-cular and macrovascular complications in diabetes? The Wis-consin Epidemiologic Study of Diabetic Retinopathy. ArchIntern Med 1997;157:650 6.14. Stratton IM, Kohner EM, Aldington SJ, et al, UKPDS Group.UKPDS 50: risk factors for incidence and progression ofretinopathy in type II diabetes over 6 years from diagnosis.Diabetologia 2001;44:156 63.15. Varma R, Torres M, Pena F, et al, Los Angeles Latino EyeStudy Group. Prevalence of diabetic retinopathy in adultLatinos: the Los Angeles Latino Eye Study. Ophthalmology2004;111:1298 306.16. West SK, Klein R, Rodriguez J, et al. Diabetes and diabeticretinopathy in a Mexican-American population: ProyectoVER. Diabetes Care 2001;24:1204 9.17. Varma R, Macias GL, Torres M, et al, Los Angeles LatinoEye Study Group. Biologic risk factors associated withdiabetic retinopathy: the Los Angeles Latino Eye Study.Ophthalmology 2007;114:1332 40.

Wong et al


18. Mitchell P, Smith W, Wang JJ, Attebo K. Prevalence ofdiabetic retinopathy in an older community: the Blue Moun-tains Eye Study. Ophthalmology 1998;105:406 11.19. Klein R, Klein BE, Moss SE, Linton KL. The Beaver DamEye Study: retinopathy in adults with newly discovered andpreviously diagnosed diabetes mellitus. Ophthalmology 1992;99:58 62.20. Al-Maskari F, El-Sadig M. Prevalence of diabetic retinopathyin the United Arab Emirates: a cross-sectional survey. BMCOphthalmol [serial online] 2007;7:11. Available at: http://www.biomedcentral.com/14712415/7/11. Accessed May 3,2008.21. Wong TY, Loon SC, Saw SM. The epidemiology of age relatedeye diseases in Asia. Br J Ophthalmol 2006;90:506 11.22. Rema M, Premkumar S, Anitha B, et al. Prevalence of diabeticretinopathy in urban India: the Chennai Urban Rural Epide-miology Study (CURES) Eye Study, I. Invest Ophthalmol VisSci 2005;46:2328 33.23. Wong TY, Klein R, Islam FM, et al. Diabetic retinopathy in amulti-ethnic cohort in the United States. Am J Ophthalmol2006;141:446 55.24. Cheung N, Wong TY. Diabetic retinopathy and systemicvascular complications. Prog Retin Eye Res 2008;27:16176.25. Cheung N, Rogers S, Couper DJ, et al. Is diabetic retinopathyan independent risk factor for ischemic stroke? Stroke 2007;38:398 401.26. Cheung N, Wang JJ, Klein R, et al. Diabetic retinopathy andthe risk of coronary heart disease: the Atherosclerosis Risk inCommunities Study. Diabetes Care 2007;30:1742 6.27. Cheung N, Wang JJ, Rogers SL, et al, ARIC (AtherosclerosisRisk In Communities) Study Investigators. Diabetic retinop-athy and risk of heart failure. J Am Coll Cardiol 2008;51:1573 8.28. Wong TY, Cheung N, Islam FM, et al. Relation of retinopathyto coronary artery calcication: the Multi-Ethnic Study ofAtherosclerosis. Am J Epidemiol 2008;167:51 8.29. Foong AW, Saw SM, Loo JL, et al. Rationale and methodol-ogy for a population-based study of eye diseases in Malaypeople: The Singapore Malay Eye Study (SiMES). Ophthal-mic Epidemiol 2007;14:2535.30. Su DH, Wong TY, Wong WL, et al, Singapore Malay EyeStudy Group. Diabetes, hyperglycemia, and central cornealthickness: the Singapore Malay Eye Study. Ophthalmology2008;115:964 8.31. Lamoreux EL, Chong E, Wang JJ, et al. Visual impairment,causes of vision loss and falls: the Singapore Malay EyeStudy. Invest Ophthalmol Vis Sci 2008;49:528 33.

Footnotes and Financial Disclosures

32. Wong TY, Chong EW, Wong WL, et al. Prevalence and causes oflow vision and blindness in an urban Malay population: the Singa-pore Malay Eye Study. Arch Ophthalmol 2008;126:10919.33. Sun C, Liew G, Wang JJ, et al. Retinal vascular caliber, bloodpressure, and cardiovascular risk factors in an Asian population:the Singapore Malay Eye Study. Invest Ophthalmol Vis Sci2008;49:1784 90.34. Shankar A, Leng C, Chia KS, et al. Association between bodymass index and chronic kidney disease in men and women:population-based study of Malay adults in Singapore. NephrolDial Transplant 2008;23:1910 8.35. Cikamatana L, Mitchell P, Rochtchina E, et al. Five-year inci-dence and progression of diabetic retinopathy in a dened olderpopulation: the Blue Mountains Eye Study. Eye 2007;21:46571.36. Eye Diseases Prevalence Research Group. The prevalence ofdiabetic retinopathy among adults in the United States. ArchOphthalmol 2004;122:552 63.37. Klein R, Klein BE, Moss SE, et al. The Wisconsin Epidemi-ologic Study of Diabetic Retinopathy. II. Prevalence and riskof diabetic retinopathy when age at diagnosis is less than 30years. Arch Ophthalmol 1984;102:520 6.38. Klein R, Klein BE, Moss SE, Cruickshanks KJ. The Wiscon-sin Epidemiologic Study of Diabetic Retinopathy. XIV. Ten-year incidence and progression of diabetic retinopathy. ArchOphthalmol 1994;112:121728.39. Klein R, Klein BE, Moss SE, et al. The Wisconsin Epidemi-ologic Study of Diabetic Retinopathy. X. Four-year incidenceand progression of diabetic retinopathy when age at diagnosisis 30 years or more. Arch Ophthalmol 1989;107:244 9.40. Klein R, Moss SE, Klein BE, et al. The Wisconsin Epidemi-ologic Study of Diabetic Retinopathy. XI. The incidence ofmacular edema. Ophthalmology 1989;96:150110.41. Teuscher A, Schnell H, Wilson PW. Incidence of diabeticretinopathy and relationship to baseline plasma glucose andblood pressure. Diabetes Care 1988;11:246 51.42. Cheung N, Wong TY. Obesity and eye diseases. Surv Oph-thalmol 2007;52:180 95.43. Dowse GK, Humphrey AR, Collins VR, et al. Prevalence andrisk factors for diabetic retinopathy in the multiethnic popu-lation of Mauritius. Am J Epidemiol 1998;147:448 57.44. Cheung N, Bluemke DA, Klein R, et al. Retinal arteriolarnarrowing and left ventricular remodeling: the Multi-EthnicStudy of Atherosclerosis. J Am Coll Cardiol 2007;50:48 55.45. Cheung N, Wong TY. Microvascular disease and cardiomy-opathy [letter]. J Card Fail 2007;13:792.

Originally received: January 14, 2008.

6

Department of Medicine, Alexandra Hospital, Singapore, Republic of

Final revision: April 27, 2008.

Singapore.

Accepted: May 6, 2008.

7

Department of Endocrinology, Singapore General Hospital, Singapore,

Available online: June 26, 2008.

Manuscript no. 2008-76.

Republic of Singapore.

1

Centre for Eye Research Australia, University of Melbourne, Melbourne,

Financial Disclosure(s):

Australia.2gapore, Republic of Singapore.3National University of Singapore, Singapore, Republic of Singapore.4

The authors have no conicts of interests or nancial disclosures related tothis article.Supported by the National Medical Research Council, Singapore (grant nos.:0796/2003, 0863/2004, and CSI/0002/2005); and the Biomedical ResearchCouncil, Singapore (grant no.: 501/1/255). Additional support was providedby the Singapore Tissue Network and the Ministry of Health, Singapore.Correspondence:

5

Department of Community, Occupational and Family Medicine, Yong

Tien Y. Wong, MD, PhD, Centre for Eye Research Australia, University of

Loo Lin School of Medicine, National University of Singapore, Singapore,Republic of Singapore.

Melbourne, 32 Gisborne Street, Victoria 3002, Australia. E-mail: [email protected].

1875Singapore Eye Research Institute, Singapore National Eye Centre, Sin-Department of Ophthalmology, Yong Loo Lin School of Medicine,Centre for Vision Research, University of Sydney, Sydney, Australia.

Documents

Diabetic Retinopathy