Ophthalmic Epidemiology, 12:143–153, 2005Copyright ©c Taylor & Francis Inc.ISSN: 0928-6586DOI: 10.1080/09286580590932734

ORIGINAL ARTICLE

Sankara Nethralaya—Diabetic RetinopathyEpidemiology and Molecular Genetic Study

(SN—DREAMS 1): Study Design andResearch Methodology

Swati Agarwal and Rajiv RamanShri Bhagwan Mahavir VitreoretinalServices, Sankara Nethralaya, Chennai,Tamil Nadu, India

Pradeep G. PaulDepartment of Epidemiology, SankaraNethralaya, Chennai, Tamil Nadu, India

Padmaja Kumari RaniShri Bhagwan Mahavir VitreoretinalServices, Sankara Nethralaya, Chennai,Tamil Nadu, India

Satagopan Uthraand Raman GayathreeDepartment of Genetics and MolecularBiology, Sankara Nethralaya, Chennai,Tamil Nadu, India

Cathy McCartyMarshfield Clinical ResearchFoundation, Marshfield,Wisconsin, USA

Govindasamy KumaramanickavelDepartment of Genetics and MolecularBiology, Sankara Nethralaya, Chennai,Tamil Nadu, India

Tarun SharmaShri Bhagwan Mahavir VitreoretinalServices, Sankara Nethralaya, Chennai,Tamil Nadu, India

ABSTRACT Purpose: To describe the methodology of the Sankara Nethralaya—Diabetic Retinopathy Epidemiology and Molecular Genetic Study (SN—DREAMS 1), an ongoing population-based study to estimate the prevalence ofdiabetes and diabetic retinopathy in urban Chennai, Tamil Nadu, South India,and also to elucidate the clinical, anthropometric, biochemical and genetic riskfactors associated with diabetic retinopathy. Methods: In this ongoing study, weanticipate recruiting a total of 5830 participants. Eligible patients, over the ageof 40 years, are enumerated using the multistage random sampling method.Demographic data, socioeconomic status, physical activity, risk of sleep apnea,dietary habits, and anthropometric measurements are collected. A detailedmedical and ocular history and a comprehensive eye examination, includingstereo fundus photographs, are taken at the base hospital. Biochemical investi-gations (total serum cholesterol, high-density lipoproteins, serum triglycerides,hemoglobin, glycosylated hemoglobin HbA1c) and genetic studies of eligiblesubjects are conducted. A computerized database is created for the records.Conclusion: The study is expected to result in an estimate of the prevalence ofdiabetes and diabetic retinopathy and a better understanding of biochemicaland genetic risk factors associated with diabetic retinopathy in an urban SouthIndian population. Worldwide, the prevalence of diabetes mellitus, in particulartype II diabetes, is rising at an alarming rate. The World Health Organization(WHO) and International Diabetes Federation (IDF) have predicted that thenumber of cases of adult-onset diabetes would more than double by 2030from the present level of 171 million to 366 million—an increase of 214%.1

In developed countries, this increase in diabetic population would be around42% and in developing countries, particularly in India, it is even higher; i.e.150%.1 In India, the prevalence of diabetes mellitus in the urban populationis around 12.1%, as reported by the national urban diabetes study2 conductedin six major cities. Studies have shown the prevalence of diabetes to be higheramong the high-income groups (25.5%) as compared to low-income groups

Received 10 June 2004Accepted 17 November 2004

Address correspondence to Dr. Tarun Sharma,MD, FRCSEd, Director, Shri Bhagwan MahavirVitreoretinal Services, Sankara Nethralaya, 18,College Road, Chennai, 600 006, Tamil Nadu,India. Fax: +91-44-28254180; E-mail:tarusharma@eth.net

143

(12.6%).3−5 The assessment of socioeconomic statuswas based on income,6,7 education,2,7 occupation2 orcaste6—which are not representative of the actual so-cioeconomic status. In the present study, however, thesample was stratified on socioeconomic scoring. Thisscoring was calculated on the basis of several parameterssuch as the residence being rented or owned, the num-ber of rooms in the house, the highest educational sta-tus, the highest salary, the highest occupation, materialpossessions (cycle, TV, audio, car, etc.) and house/landvalue. To the best of our knowledge, this kind of com-prehensive socioeconomic scoring has not been donebefore for prevalence studies on diabetic retinopathyin the general population.

KEYWORDS Diabetic retinopathy; diabetes mellitus; epi-demiology; genetics; prevalence; risk factors; India

INTRODUCTIONThe prevalence of diabetic retinopathy in a diabetic

population ranges from 10.2 to 48%.8−14 In India,the prevalence of diabetic retinopathy among diabeticsranges from 22.4 to 34.1%;15−17 since this prevalencehas been estimated primarily in self-reported diabeticsor in specialist diabetic centers, it does not truly reflectthe prevalence of diabetic retinopathy in previouslydiagnosed and newly diagnosed diabetics in the popu-lation. The present study addresses this issue by includ-ing both newly diagnosed and previously diagnoseddiabetics.

Genetic factors play an important role in the patho-genesis of diabetic retinopathy. In India, until now, nopopulation-based studies have evaluated the genetic fac-tors influencing diabetic retinopathy. The importantcandidates are the vascular endothelial growth factor(VEGF) gene,18 the receptor for advanced glycosyla-tion end products (RAGE) gene,19,20 the protein kinaseC (PKC)–β gene,21,22 and the apolipoprotein E (ApoE) gene.23 In our previous work in the Asian Indianpopulation, we found that the Z-2 allele of the aldosereductase gene24 is a risk factor for developing diabeticretinopathy, and also observed that Gly82Ser polymor-phism in the RAGE gene20 was a protective allele forretinopathy whereas the tumor necrosis factor 111 bpallele of dinucleotide microsatellite repeat25 was associ-ated with proliferative diabetic retinopathy. Our studyon the inducible nitric oxide synthase gene26 showed

that a 210 bp pentanucleotide repeat 2.5 kb upstreamfrom the transcription site was a low risk allele for de-veloping retinopathy.

The present population-based study has been de-signed using a multistage random sampling techniqueto answer the following questions:

What is the prevalence of diabetes mellitus in an urbanpopulation of Chennai, Tamil Nadu, India?

What is the prevalence of diabetic retinopathy in thiswell-defined population of diabetics, both newly di-agnosed and previously diagnosed?

What are the risk factors—clinical, sociodemographic,anthropomorphic, and biochemical—associated withdiabetic retinopathy?

What sequence variants or polymorphisms in theRAGE, VEGF, PKC–β, and ApoE genes are asso-ciated with diabetic retinopathy?

This paper will elucidate the research methodologydesigned for SN-DREAMS.

MATERIALS AND METHODSThe study procedure is divided into five phases (see

Fig. 1):

Phase 1: Epidemiological field workPhase 2: Training program, quality control, pilot study,

definitionsPhase 3: Medical history questionnairePhase 4: Ophthalmic examination, fundus photogra-

phy, diabetic retinopathy classificationPhase 5: Biochemical investigations and genetic study

Phase 1: Epidemiological FieldworkThe epidemiological fieldwork, commenced in

October 2003, is expected to be completed byDecember 2005. The clinical examination, biochem-ical investigations and genetic study, commenced inJanuary 2004, will be completed by April 2006. TheInstitutional Review Board at Sankara Nethralaya hasapproved this project.

Study Design

SN-DREAMS is a population-based cross-sectionalsurvey, which measures the prevalence of diabetes anddiabetic retinopathy and estimates the risk or protectivefactors associated with diabetic retinopathy.

S. Agarwal et al. 144

FIGURE 1 Study Design. RAGE = Receptor for Age-related End Products; VEGF = Vascular Endothelial Growth Factor; PKC-β =Protein kinase C-β.

Sample Size Estimation

In India, Dandona et al.15 estimated the prevalenceof diabetic retinopathy to be 1.78% in a population-based sample of adults aged 30 years or older, andNarendra et al.16 reported a prevalence of 1.4% in self-reported diabetics aged 50 years or above. Based onthese published studies, it is appropriate to assume thatthe prevalence of diabetic retinopathy in the generalpopulation may range from 1 to 2%.

In our study, the computed sample size is 5830. Thisestimation is based on the following assumptions: the

prevalence of diabetic retinopathy in the general pop-ulation is assumed to be 1.3%, with a relative preci-sion of 25%, a drop-out rate of 20%, and a designeffect of 2. The sample size for the prevalence sur-vey was calculated by using the formula 4PQ/d2 whereP is the expected prevalence; Q = 1-P and d is theprecision.27−29

Study Area

Chennai (formerly Madras), the fourth largest cityin India, is located on the Coromandel Coast of the

145 SN-Diabetic Retinopathy Study

TABLE 1 Demographic Data on the Study Area

Parameters Chennai District

Area 174 sq. kmPopulation 4.3 millionMale 2.2 millionFemale 2.1 millionLiteracy rate 76.8%% Population >40 yrs 28%

Major religions:Hindus: 3.6 million (82.3%)Muslims: 379,206 (8.7%)Christians: 331,261 (7.6%)Others: 51791 (1.2%)Religion not stated: 8031 (0.2%)

Bay of Bengal. Chennai city has an area of 174 sq. km,accounting for 0.13% of the State’s total area of1,300,058 sq. km; the entire area of the district hasbeen classified as urban. It has a population of around4.3 million with 2.2 million males and 2.1 million fe-males according to the census of 2001; the literacy rateis 76.8% (Table 1).

FIGURE 2 Map of Chennai showing the divisions—the shadedareas represent study divisions.

Sampling Method

Chennai city is divided into 10 corporation zones of155 divisions (Fig. 2). The sampling is based on multi-stage systematic random sampling. Sampling is done intwo stages: selection of divisions and selection of studysubjects. Selection of divisions is done using computer-generated random numbers; of 155 divisions, 10 are se-lected ensuring that one division per one corporate zoneis represented in the sample. Eligible study subjects arerandomly selected from each division. To meet the tar-get, 600 individuals are enumerated for each division(a total of 6000 in 10 zones). This sample is thus trulyrepresentative of urban Chennai, Tamil Nadu, India.

The sampling is done to ensure that data is collectedfrom all socioeconomic groups. Socioeconomic data30

are collected from every 5th household on both sidesof the street. Using a scoring system, each street is cat-egorized as being a low (score 0–14), middle (score 15–28) or high (score 29–42) socioeconomic status street.To obtain a sample of 600 from each division and dis-tribute it evenly based on socioeconomic criteria, thepopulation proportion to size sample (PPS) from eachsocioeconomic stratum is calculated.

Household Enumeration

Family members living on the same premises andsharing a common kitchen are defined as being mem-bers of one household. A door-to-door survey of all thehouseholds on the right side of the street is conductedin the selected division until a number of 600 subjectsis reached. The household data sheet contains detailsof demography, educational qualification, occupation,residential status, and diabetic status. Each subject isassigned a unique eleven-digit identification number.For the population, the first two digits indicate thedivision, the third and the fourth represent the streetnumber, the next three represent the door number, thefollowing two represent the household number, and thelast two denote the individual serial number.

Eligibility Criteria

Individuals aged 40 years or above or turning 40 inthe current year and residing for a minimum of sixmonths at the same residence.

Exclusion Criteria

Individuals residing for a period of less than sixmonths at the same residence, temporary residents

S. Agarwal et al. 146

(whose permanent residence is elsewhere), a resi-dent who dies after the enumeration but prior toexamination, and those who cannot be contacted af-ter five attempts by the social worker at their residence.Very old or invalid persons (blind subjects are not con-sidered as invalid) who cannot be transported to theexamination center are also excluded.

Refusals

When the occupants of the household refuse to pro-vide any information or to comply with the examina-tion, it is taken as a refusal. A maximum of five attemptsare made to convince them before considering them asdropouts. The strategy to minimize refusals includes in-forming all the study participants about the study bene-fits plus sending reminders via the telephone or makinga personal visit by the social worker.

Estimation of Fasting Blood Glucose

The team visits the selected households a day in ad-vance to request eligible individuals to observe a min-imum of 8 hours of overnight fasting prior to estima-tion of the fasting capillary blood glucose levels thenext morning. Using Accutrend Alpha, blood glucoseestimation is done on a sample of capillary blood ob-tained by finger-prick (glucose oxidase method). In ac-cordance with the ADA criteria,31 known diabetics andnew asymptomatic subjects with a fasting blood glu-cose ≥110 mg/dl (provisional diabetics) are selected.For patients who forget the fasting instructions and fordropouts, the fasting blood glucose is estimated on Sun-days or at their convenience.

All known diabetics and provisional diabetics fromthe field area are given appointments for furtherevaluation at the base hospital. On the day of theexamination, the subjects are instructed to assembleat a specific place with their prescriptions or medicinethat they are taking for ophthalmic or systemic con-ditions; the social worker accompanies the subjects inthe project vehicle to the hospital. Once the compre-hensive evaluation is over, the subjects are transportedback to their residences.

Phase 2: Training Program, QualityControl, Pilot Study, Definitions

Training Program

The epidemiology team of Sankara Nethralaya pro-vided intensive training on a one-to-one basis to all the

team members. This lasted for 7 days with 8 hours aday of training sessions. The aim was to ensure thateach member of the team was well trained in doinga household survey, enumeration, and filling out thestudy data sheet. The trainees were instructed to usethe BP apparatus and glucometer for estimating fast-ing capillary blood glucose. The main objective was toavoid bias or errors in any of the procedures employed.Each trainee was evaluated individually and allowed toparticipate in the study only after he or she displayedminimum error rates for the tasks involved in the study.

Quality Control

In order to ensure accurate and reliable data, acomprehensive instruction manual has been prepared.A start-up training session helped us to standardizeall the examination and diagnostic procedures. Theglucometer32 is calibrated every day and its repro-ducibility is assessed by measuring the blood glucosefor the same patient six times and also with two ma-chines. A similar procedure is undertaken for the sphyg-momanometer. The scale for measuring the weight iscalibrated with a known weight once a week. The col-lected data is scrutinized manually before its entry intothe computer.

Pilot Study

A pilot study was conducted taking 139 patients fromthe first division (Chintradipet); this was done to iden-tify the initial practical difficulties in carrying out thestudy. The first step was the selection of the partici-pants. Next, informed consent was obtained after thestudy had been explained to each participant and theinterviewer was certain that the participant understoodand accepted the contents. The subjects underwent theentire sequence of examination and laboratory proce-dures as planned for the actual study.

During the pilot study, inter- and intraobserver vari-ations for the photographic classification of diabeticretinopathy were compared (in order to minimize bias);kappa was 0.83 and 1.0, respectively. A Kappa value of>0.8 suggests good agreement between the observers.

A time-motion study was done to determine the timeneeded for a complete examination. The study revealedthat it took 2 hours and 45 minutes for the collection ofall the data, which included medical history, question-naires and anthropometric measurements (45 minutes);ophthalmic work-up with corneal examination(1 hour); dilatation, fundus examination, and fundus

147 SN-Diabetic Retinopathy Study

photography (30 minutes); vibration perceptionthreshold (VPT) testing and genetic work-up(30 minutes). The pilot study concluded that there wasno questionnaire fatigue and a good response rate, creat-ing the necessary confidence to proceed with the study.

Overview

All known diabetics and provisional diabetics (firstFBS ≥110 mg/dl) are invited to the diabetic retinopathyproject site (base hospital) of Sankara Nethralaya forfurther evaluation; provisional diabetics are instructedto come fasting on the day of examination. On arrivalat the reception, the subjects are registered and issuedthe previously designed eleven-digit ID number. A newfour-digit hospital diabetic retinopathy (DR) number isgiven. The subjects then undergo blood collection bythe laboratory technician. Various definitions used inthe study are summarized in Table 2.

Phase 3: Medical History, GeneralPhysical Examinationand Questionnaires

Medical and Ocular History

In this phase, a detailed questionnaire is administeredregarding the medical history and a general physicalexamination is given. The data in the medical historyinclude duration33 and treatment of diabetes or hyper-tension, a family history of diabetes, coronary arterydisease, symptoms related to diabetic nephropathy orneuropathy (tingling, numbness, foot ulcers and ampu-tated toes or foot), the presence of an ear lobe crease,34

smoking or alcohol consumption,35 pregnancy, and du-ration of total sleep.36 The ocular history includes de-tails of the first and last eye examination, present orpast ocular complaints, and laser treatment or ocularsurgery.

TABLE 2 Definitions

Provisional diabetics New asymptomatic individual with a first fasting blood glucose level ≥110 mg/dl (Accutrendalpha).

Known diabetics Diagnosis of diabetes made by a medical practitioner, or patient using hypoglycemicmedication, either oral or insulin or both.

Newly diagnosed diabetics Fasting blood glucose level ≥110 mg/dl on two separate days; provisional diabetics wereretested at the base hospital by the laboratory method.

Duration of diabetes Time interval between the date of diagnosis of diabetes (as made by a diabetologist or whenthe antidiabetic treatment started) and the date of eye examination.

Hypertension If the systolic BP is ≥140 mm Hg or the diastolic BP is ≥ 90 mm Hg or the patient is onantihypertensive treatment.

General Physical Examination

This includes measurement of height and weight, af-ter which the body mass index (BMI)16,35 is calculatedusing the formula: weight (kg)/height (m2). Based onthe BMI, individuals are classified as lean (male <20,female <19), normal (male 20–25, female 19–24), over-weight (male 25–30, female 24–29) or obese (male >30,female >29).

The waist and hip ratio (WHR)37 is calculated bydividing the waist circumference (cm) by the hip cir-cumference (cm). The neck circumference38 and thighlength39 are also measured. The resting heart rate40,41

per minute is taken after making the patient sit for atleast five minutes. The blood pressure is recorded in thesitting position on the right arm; two readings are taken5 minutes apart and the mean of the two is taken as theblood pressure.

The Epworth sleepiness score42 is calculated by elic-iting the chance of dozing off (never, slight, moderateor high) during the day while reading, watching TV,sitting inactive in a public place, traveling in a car forhalf an hour without a break, lying down in the after-noon and so on. Sleep apnea42 is considered if any ofthe following two variables are present: obesity (BMImale >30, female >29) or hypertension (>140/90),a history of snoring and an Epworth sleepinessscore ≥5.

Questionnaires

Physical activity is assessed by means of a simplequestionnaire,43 which has been validated in the SouthIndian population. Socioeconomic scores are notedon the basis of the questionnaire used in the firstphase.

A knowledge, aptitude, and practice (KAP) ques-tionnaire about diabetic retinopathy is filled in foreach individual, as well as a diet questionnaire. For all

S. Agarwal et al. 148

patients, peripheral neuropathy foot testing is done bymeasuring the vibration perception threshold.44

Phase 4: Ophthalmic Examination,Fundus Photography, Diabetic

Retinopathy ClassificationVisual Acuity and Refraction

The modified ETDRS chart (Light House Low Vi-sion Products, New York, NY, USA) is used to test thedistance visual acuity; and for those who cannot readthe English alphabet, Landolt’s ring test is used. If thevisual acuity is less than 4/4 (logMAR 0.0), the pinholevisual acuity is assessed.

Objective refraction is performed with a streakretinoscope (Beta 200, Heine, Germany) followed bysubjective refraction. If the subject is unable to readthe 4/40 (logMAR 1.0) line, vision is checked at onemeter. If he or she is still unable to identify any ofthe largest optotypes, perception of hand movementsis observed. If hand movements cannot be identified,the examiner checks for perception of light, which isrecorded as present or absent.

External Examination

External examination is performed using a handheldflashlight. The face and eyes are examined for the pres-ence of strabismus, extraocular movement abnormali-ties or any other gross abnormality.

Corneal Examination

This consists of corneal pachymetry, Schirmer’s test,specular microscopy, and Rose Bengal and fluoresceinstaining. It is believed that corneal thickness plus en-dothelial cell density, pleomorphism and polymegath-ism are positively correlated with the type, duration andseverity of diabetes; as is the case for tear film and ocularsurface dysfunction.45−47 However, these relationshipshave not been substantiated in the population-basedstudies.

Corneal pachymetry48 (Alcon ultrasound pachy-meter) is performed to measure the corneal thickness.Measurements are taken for the central and inferiorperipheral cornea. If measurements of the inferior pe-ripheral cornea are not possible, a temporal peripheralreading is taken. An average of three readings are takenfor each eye.

Schirmer’s test49,50 is done to assess the aqueouscomponent of the tears. Whatman’s no. 41 filter pa-

per is bent at the notch and inserted in the fornix at thejunction of the medial 2/3 and lateral 1/3 of the lowereyelid, into each eye in succession with the shortest pos-sible interval; care is taken not to touch the cornea. Thepatient is allowed to keep the eye either closed or open.At the end of 5 minutes, the length of the strip that hasbecome wet is measured; more than 10 mm of wettingis considered normal.

Corneal specular microscopy51−53 is performed toassess the corneal endothelial status. With the patientsitting and the head position adjusted to obtain a com-plete and well-focused corneal reflex on the center ofthe cornea, the record button is pressed; if the centralcorneal pictures are not good, the procedure is repeatedin any of the peripheral quadrants. The data are ana-lyzed with regard to cell density, percentage of hexago-nality, coefficient of variance, and number of cells in-cluding the standard deviation.

Rose Bengal and fluorescein staining54−56 is used forthe diagnosis of ocular surface diseases, especially forthe screening of dry eye. A drop of 1% Rose Bengal isplaced on a strip of fluorescein paper which is instilledinto the cul-de-sac of the patient’s eye; separate strips areused for each eye. The patient thereafter blinks naturallyand the staining pattern is studied after 1–2 minutes (toallow excess stain in the precorneal film to disappear)by slit-lamp biomicroscopy with a red-free filter. Thecornea, the nasal part of the conjunctiva and the tem-poral conjunctiva are scored separately. A score of 1 isgiven if a few separated spots of stain are seen, a scoreof 2 if multiple but separated spots appear, and a scoreof 3 if there are confluent spots. Thus, the total scoreranges from 0 to 9.

Using a blue filter, fluorescein staining of the corneais also studied by the slit-lamp. For documentation pur-poses, the cornea is divided into 3 zones: superior, mid-dle and inferior 1/3. The grading of the staining is thesame as that used for Rose Bengal.

Slit-Lamp Examination

The Zeiss SL 130 (Carl Zeiss, Jena, Germany) slit-lamp is used. Using a moderately wide beam, the eye-lids, margins, lashes, canthi, and puncta are systemati-cally examined, followed by the palpebral and bulbarconjunctiva, sclera and cornea. Then, using a narrowparallelopiped beam, the cornea, anterior chamber andiris are examined for abnormality. Grading of periph-eral anterior chamber depth is done according to theVan Herick grading.57 The iris and pupillary margins

149 SN-Diabetic Retinopathy Study

are examined under high magnification to look for irisneovascularization.

Applanation Tonometry 58

Using the Goldmann applanation tonometer (ZeissAT 030 Applanation Tonometer, Carl Zeiss, Jena,Germany), the intraocular pressure (IOP) is measuredin both eyes; 0.5% proparacaine eyedrops are used fortopical anesthesia, and a 2% fluorescein strip to stainthe tear film.

Gonioscopy

Gonioscopy is performed in dim ambient illumina-tion with a shortened slit that does not fall on the pupil.A Sussmann-type 4-mirror handheld gonioscope (VolkOptical Inc, Mentor, Ohio, USA) is used. Gonioscopyis performed only on those patients who have an ele-vated intraocular pressure, iris neovascularization or ashallow anterior chamber (Van Herrick’s grading ≤2).

The angle is graded according to the Shaffergrading;59 the peripheral iris contours, degree of tra-becular meshwork pigmentation and angle neovascu-larization are also recorded. In eyes with an occludableangle, laser iridotomy is performed before dilatation;the rest of examination is done at a later date.

Grading of Lens Opacities

The subject’s pupils are dilated with 5% phenyl-ephrine and 1% tropicamide eyedrops; if phenyle-phrine is contraindicated, 1% cyclopentolate eyedropsare used.

Grading of lens opacities is performed using the LensOpacities Classification System60 (LOCS chart III, LeoT. Chylack, Harvard Medical School, Boston, MA).Lenticular opacities are graded by comparison with thestandard set of photographs, which are retroilluminatedby mounting on a light box.

With the slit beam at a 45◦ angle, the slit height andwidth are adjusted to approximate the overall brightnessof the corneal image and anterior subcapsular zone tonuclear color/opalescence standard N1; 0◦ brightness isallowed to visualize all the lens opacities without caus-ing discomfort to the patient. Keeping the slit width at0.2 mm, nuclear opalescence (NO) and nuclear color(NC) are graded by comparing the slit-lamp examina-tion image with the nuclear standards NO1 to NO6and NC1 to NC6.

Cortical cataract (C) and posterior subcapsularcataract (P) are graded by examining the opacity in

retroillumination images (0◦ angle) focused either an-teriorly (at the iris plane) or posteriorly (at the planeof the posterior capsule) and comparing these exami-nation pictures to the cortical standards C1 to C5 andposterior subcapsular standards P1 to P5.

Fundus Examination and Clinical Gradingof Diabetic Retinopathy

The binocular indirect ophthalmoscope (Keeler In-struments Inc., Pennsylvania, USA) and +20 D lens(Nikon) are used to examine the fundus. Diabeticretinopathy is clinically graded according to the newdisease severity scale given by the American Academyof Ophthalmology;61 the macular area is specificallyexamined with a +78 D lens (Nikon) to diagnose clini-cally significant macular edema as defined by the EarlyTreatment Diabetic Retinopathy Study (ETDRS).62

Fundus Photography Using Stereo-Pictures

Irrespective of the presence or absence of diabeticretinopathy, 45◦ four-field stereoscopic digital pho-tographs (posterior pole, nasal field, superior and in-ferior) are taken for all subjects with a Carl Zeiss fun-dus camera (Visucamlite, Jena, Germany). However, forthose who are found to have diabetic retinopathy onclinical examination, additional 30◦ seven-field stereo-digital pairs are taken.

Evaluation of Photographic Gradingand Quality

The photographic grading and quality are assessedusing the Visupac digital image archiving system. The“Screenscope” (Berezin Stereo Photography Products,Mission Viejo, CA, USA), a stereo viewer that can befixed on a computer monitor, is used to examine thestereo pairs. Photographs of each eye are reviewed andgiven grades for overall quality. Field definition andimage clarity are graded as inadequate for reading orgrading, adequate (sufficient visualization of disc, mac-ula, and vessels to grade), and good (small vessels orretinal details visible across 90% of the image).

Photographic Grading of DiabeticRetinopathy

The modified classification of diabetic retinopathybased on the degrees of retinopathy used by Kleinet al.63 is used (Table 3); digital photographs are as-sessed and graded by three independent observers (ex-perienced retinal specialists) in a masked fashion. The

S. Agarwal et al. 150

TABLE 3 Diabetic Retinopathy is Graded as Follows63

No DR No abnormalityMild NPDR Only microaneurysmModerate NPDR More than mild, but less than severeSevere NPDR Any of the following: 20 or more

intraretinal hemorrhages in 4quadrants, venous beading in >2quadrants or intraretinalneovascularization in 1 quadrant

Proliferative DR One or more of the following:neovascularization or preretinalor vitreous hemorrhage

photographs are graded against the standard pho-tographs of the ETDRS grading system for severity ofretinopathy. In case of any disparity in the findings, thefirst grader consults the second grader before reachingconclusions. If the two graders do not agree, the thirdgrader’s opinion is sought.

Phase 5: Biochemicaland Genetic Studies

Biochemical Investigations

In case of patients with provisional diabetes, confir-mation is done by re-estimation of the fasting blood glu-cose by enzymatic assay; glucose is oxidized by glucoseoxidase to produce gluconate and hydrogen peroxide,which is detected photometrically.

Biochemical analysis is done on the Merck MicroLab 120 semi-automated analyzer. Total serum choles-terol (CHOD-POD method), high-density lipopro-teins (CHOD-POD method after protein precipita-tion), serum triglycerides (CHOD-POD), hemoglobin(calorimetric hemoglobinometer), packed cell volume(capillary method) and the lycosylated hemoglobinfraction (Bio-Rad DiaSTAT HbA1c Reagent Kit) areestimated. Microalbuminuria estimation is done by asemi-quantitative procedure (Clintek 50 Bayer UrineAnalyzer) with the first morning urine sample.

Genetic Analysis

From the blood sample collected, the DNA isextracted from the heparinized blood by the phe-nol/chloroform method. The extracted DNA is quan-tified spectrophotometrically, used for the polymerasechain reaction and amplified using specific primers. Fur-ther screening of polymorphisms and sequence vari-ants and their associations with the disease (diabetic

retinopathy) in the candidate genes is done by observ-ing the restriction fragment length polymorphism anddirect sequencing (VEGF, PKC-β, receptors for AGEPand Apo E). The presence of a risk allele in the genesfor aldose reductase and inducible nitric oxide synthaseand their association with the disease are studied. Theinvestigations on the patients are performed in accor-dance with the Declaration of Helsinki.64

Data ManagementData Collection Forms

Identification data, history, the results of the medicaland general physical examination along with the resultsof all the biochemical tests are entered onto the studydata sheet designed in accordance with the OmniEx-tract ICR service package (Newgen Software Technolo-gies Ltd). After the examination is over and before thepatient leaves the base hospital, the records are checkedfor completeness.

Data Entry and Processing

All of the data are fed into the computer using theOmni Extract ICR service package and scanner. Twodata operators validate the data entry.

Statistical Analysis

The prevalence of diabetes and diabetic retinopathywill be expressed as percentages with the 95% confi-dence interval. Logistic regression analysis will be doneto elucidate the association between risk factors and di-abetic retinopathy. Odds ratios with 95% confidenceintervals will be calculated for the studied variables.

Major ChallengesAs we plan to have an equal share of individuals from

all socioeconomic strata, poor compliance with the ex-amination for the study is expected particularly fromthose in the higher strata. The solution to this chal-lenge is to provide some flexibility in scheduling theireye examination based on their convenience and avail-ability. Other challenges include refusals and dropoutsand uniformity in the data collection. The issue of re-fusals and dropouts has been addressed by showing thebenefit of this study to all the participants prior to theirenrollment. To ensure uniformity in examination anddata collection, the team underwent a comprehensivetraining program prior to commencement of the study.

151 SN-Diabetic Retinopathy Study

LimitationsThe present study, being a cross-sectional survey, has

its inherent limitations. As the sample size was calcu-lated for the estimation of the prevalence of diabeticretinopathy in the general population, the power toelucidate associated risk factors in the subgroup anal-ysis may be inadequate. Secondly, this study is unableto provide data on the progression of diabetic retinopa-thy over time, as no follow-up is envisaged.

CONCLUSIONSN-DREAMS is an ongoing population-based study.

Data collection and analysis are expected to last forabout three years from commencement. The project isdesigned to provide valuable information on the preva-lence of diabetes and diabetic retinopathy in a repre-sentative sample of an urban population and the riskfactors associated with diabetic retinopathy. Additionalinsights will be obtained into the influence of geneticaspects on diabetic retinopathy. The study results willhelp plan service delivery in this rapidly spreading epi-demic of diabetes.

ACKNOWLEDGEMENTSWe are grateful to the RD Tata Trust, Mumbai, India

for financial support of the study.

THE SN-DREAMS GROUPPrincipal Investigator: Dr. Tarun SharmaAssociate Investigator: Dr. G. KumaramanickavelOphthalmologists: Dr. Rajiv Raman, Dr. Swati Agarwal,

Dr. R. Padmaja KumariResearch fellows: Dr. Vikrant Sharma, Dr. Sachin

MahuliEpidemiologist: Dr. Pradeep G. PaulOptometrists: Mr. D. Senthil Kumar, Mr. M.

GouthamanPhotographer: Mr. M. S. KrishnaField Operation Incharge: Mr. T. ArokiasamySocial workers: Mr. J. Singaraj, Mr. E. Chandrasekaran,

Mr. K. JayandrakumarProject Coordinator: Mr. K. Senthil KumaranSecretarial assistance: Ms. P. Sabitha, Ms. R. Vimala,

Mr. P. GnanamoorthyGenetics: Ms. S. Uthra, Ms. L. PraveenaAnimators: Mr. R. Alexander, Mr. M. Thilak

REFERENCES[1] Wild S, Roglic G, Green A, et al. Global prevalence of diabetes.

Diabetes Care. 2004;27:1047–1053.[2] Ramachandran A, Snehalatha C, Kapur A, et al. High prevalence of

diabetes and impaired glucose tolerance in India: National UrbanDiabetes Survey. Diabetologia. 2001;44:1094–1101.

[3] Mohan V, Shanthirani CS, Deepa R, et al. Intra-urban differencesin the prevalence of the metabolic syndrome in southern India—the Chennai Urban Population study (CUPS No. 4). Diabet Med.2001;18:280–287.

[4] Ramachandran A, Snehalatha C, Vijay V, et al. Impact of poverty onthe prevalence of diabetes and its complications in urban southernIndia. Diabet Med. 2002;19:130–135.

[5] Dudeja V, Misra A, Pandey RM, et al. BMI does not accuratelypredict overweight in Asian Indians in northern India. Br J Nutrit.2001;86:105–112.

[6] Dandona R, Dandona L, Naduvilath TJ, et al. Design of a population-based study of visual impairment in India: The Andhra Pradesh Eyedisease study. Indian J Ophthalmol. 1997;45:251–257.

[7] Deepa M, Pradeepa R, Rema M, et al. The Chennai Urban Rural Epi-demiology Study (CURES)—Study design and methodology (Urbancomponent) (CURES—1). J Assoc Physicians India. 2003;51:863–870.

[8] Rajala U, Laakso M, Qiao Q, et al. Prevalence of retinopathy in peo-ple with diabetes, impaired glucose tolerance, and normal glucosetolerance. Diabetes Care. 1998;21(10):1664–1669.

[9] Tapp RJ, Shaw JE, Harper CA, et al. The Australian DiabetesStudy Group. The prevalence of and factors associated with di-abetic retinopathy in the Australian population. Diabetes Care.2003;26(6):1731–1737.

[10] The Eye Diseases Prevalence Research Group. The prevalence of dia-betic retinopathy among adults in the United States. Arch Ophthal-mol. 2004;122:552–563.

[11] Hove MN, Kristensen JK, Lauritzen T, et al. The prevalence ofretinopathy in an unselected population of type 2 diabetes pa-tients from Arhus County, Denmark. Acta Ophthalmol Scand Online.2004;10.1111/j.1600-0420.2004.00270.x

[12] Broadbent DM, Scott JA, Vora JP, et al. Prevalence of diabetic eyedisease in an inner city population: The Liverpool Diabetic Eye Study.Eye. 1999;13(Pt 2):160–165.

[13] Khandekar R, Al Lawatii J, Mohammed AJ, et al. Diabetic retinopathyin Oman: a hospital based study. Br J Ophthalmol. 2003;87(9):1061–1064.

[14] West SK, Klein R, Rodriguez J, Munoz B, et al. Diabetes and dia-betic retinopathy in a Mexican-American population: Proyecto VER.Diabetes Care. 2001;24(7):1204–1209.

[15] Dandona L, Dandona R, Naduvilath TJ, et al. Population based as-sessment of diabetic retinopathy in an urban population in southernIndia. Br J Ophthalmol. 1999;83:937–940.

[16] Narendra V, John RK, Raghuram A, et al. Diabetic retinopathy amongself-reported diabetics in southern India: a population based assess-ment. Br J Ophthalmol. 2002;86:1014–1018.

[17] Rema M, Deepa R, Mohan V. Prevalence of retinopathy at diagnosisamong type 2 diabetic patients attending a diabetic centre in southIndia. Br J Ophthalmol. 2000;84:1058–1060.

[18] Awata T, Inoue K, Kurihara S, et al. A common polymorphism in the5′-untranslated region of the VEGF gene is associated with diabeticretinopathy in type 2 diabetes. Diabetes. 2002;51:1635–1639.

[19] Warpeha KM, Chakravarthy U. Molecular genetics of microvasculardisease in diabetic retinopathy. Eye. 2003;17(3):305–311.

[20] Kumaramanickavel G, Ramprasad VL, Sripriya S, et al. Association ofGly82Ser polymorphism in the RAGE gene with diabetic retinopathyin type II diabetic Asian Indian patients. J Diabetes Complications.2002;16(6):391–394.

[21] Ishii H, Jirousek MR, Koya D, et al. Amelioration of vasculardysfunctions in diabetic rats by an oral PKCβ inhibitor. Science.1996;272:728–731.

S. Agarwal et al. 152

[22] Lloyd PA. The potential role of PKC-β in diabetic retinopathy andmacular edema. Surv Ophthalmol. 2002;47(Suppl 2):S263–269.

[23] Shcherbak NS. Apolipoprotein E gene polymorphism is not a strongrisk factor for diabetic nephropathy and retinopathy in Type I dia-betes: case control study. BMC Med Gene. 2001;2:8.

[24] Kumaramanickavel G, Sripriya S, Ramprasad VL, et al. Z-2 aldosereductase allele and diabetic retinopathy in India; Ophthalmic Genet.2003;24(1):41–48.

[25] Kumaramanickavel G, Sripriya S, Vellanki RN, et al. Tumor necro-sis factor allelic polymorphism with diabetic retinopathy in India.Diabetes Res Clin Pract. 2001;54(2):89–94.

[26] Kumaramanickavel G, Sripriya S, Vellanki RN, et al. Inducible nitricoxide synthase gene and diabetic retinopathy in Asian Indian pa-tients. Clin Genet. 2002;61(5):344–348.

[27] Lwanga SK, Lemeshow S. Sample size determination in health stud-ies; a practical manual. Geneva: World Health Organisation, 1991.

[28] Bennett S, Woods T, Lyange WM, Smith DL. A simplified generalmethod for cluster-sample surveys of health in developing countries.World Health Stat Q. 1991;44:98–106.

[29] Carlin JB, Hocking J. Design of cross-sectional surveys using clustersampling: an overview with Australian case studies. Aust N Z J PublicHealth. 1999;23:546–551.

[30] Oakes JM, Rossi PH. The measurements of SES in health research:current practice and steps towards a new approach. Soc Sci Med.2003;56:769–784.

[31] Diagnostic criteria for diabetes mellitus. Diabetes Care.2003;26(Suppl 1):5S.

[32] Guun BB, Kristensen KN, Thue G, et al. Standardized evaluation ofinstruments for self-monitoring of blood glucose by patients and atechnologists. Clin Chem. 2004;50:1068–1071.

[33] Klein R, Klein BEK, Moss SE, et al. The Wisconsin Epidemiologic studyof Diabetic Retinopathy IV. Ten-year incidence and progression ofdiabetic retinopathy. Arch Ophthalmol. 1994;112:1217–1228.

[34] David TM, Balme M, Jackson D, et al. The diagonal ear lobe crease(Frank’s sign) is not associated with coronary artery disease orretinopathy in type 2 diabetes: the Fremantle Diabetes Study. AustN Z J Med. 2000;30(5):573–577.

[35] Kohner EM, Aldington SJ, et al. United Kingdom Prospective Di-abetes Study, 30. Diabetic retinopathy at diagnosis of non-insulindependent diabetes mellitus and associated risk factors. Arch Oph-thalmol. 1998;116:297–303.

[36] Get enough sleep and avoid diabetes. Diabetes Care. 2003;26:380–384.

[37] Ramachandran A, Snehalatha C, Vijay V, et al. Clustering of car-diovascular risk factors in urban Asian Indians. Diabetes Care.1998;21(6):967–971.

[38] Khawaja IT, Phillips BA. Obstructive sleep apnea: diagnosis and treat-ment. Hosp Med. 1998;34:33–36.

[39] Kao WHL, Baptiste-Roberts K, Bandeen-Roche K, et al. Short thighsof 15.1 inches linked to greater risk of diabetes. American HeartAssociation Conference, March 7, 2003.

[40] Klein R, et al. The Wisconsin Epidemiologic study of DiabeticRetinopathy II. Prevalence and risk of diabetic retinopathy whenage at diagnosis is less than 30 years. III. Prevalence and risk of di-abetic retinopathy when age at diagnosis is 30 or more years. ArchOphthalmol. 1984;102:520–532.

[41] Wong TY, Moss SE, Klein R, et al. Is the pulse rate useful in assess-ing risk of diabetic retinopathy and macular oedema? The Wiscon-sin Epidemiological Study of Diabetic Retinopathy. Br J Ophthalmol.2001;85:925–927.

[42] John MW. A new method for measuring daytime sleepiness: TheEpworth Sleepiness Scale. Sleep. 1991;14(6):540–545.

[43] Bharathi AV, Vaz M. The construct of a simple clinic questionnaireto assess physical activity and its relative validity. Indian Heart J.2000;52:601–603.

[44] Armstrong DG, Lavery LA, Vela SA, et al. Choosing a practical screen-ing instrument to identify patients at risk for diabetic foot ulceration.Arch Internal Med. 1998;158(3):289–292.

[45] Roszkowska AM, Tringali CG, Colosi P, Squeri CA, Ferreri G. Cornealendothelium evaluation in type I and type II diabetes mellitus. Oph-thalmologica. 1999;213(4):258–261.

[46] Siribunkum J, Kosrirukvongs P, Singalavanija A. Corneal abnormali-ties in diabetes. J Med Assoc Thailand. 2001;84(8):1075–1083.

[47] Ozdemir M, Buyukbese MA, Cetinkaya A, Ozdemir G. Risk factors forocular surface disorders in patients with diabetes mellitus. DiabetesRes Clin Pract. 2003;59(3):195–199.

[48] Salz JJ, et al. Evaluation and comparison of sources of variability inthe measurement of corneal thickness with ultrasonic and opticalpachymeters. Ophthalmic Surg. 1983;14:750–754.

[49] Schirmer O. Studien zur Physiologie und Pathologie der Tranenab-sonderung und Tranenabfuhr. Graefes Arch Clin Exp Ophthalmol.1903;56:197.

[50] Lucca JA, Nunez JN, Farris RL. A comparison of diagnostic tests forkeratoconjunctivitis sicca: lactoplate, Schirmer and tear osmolarity.CLAO J. 1990;16:109–112.

[51] Laing R, Sandstrom M, Leibowitz H. Clinical specular microscopy. I.Optical principles. Arch Ophthalmol. 1979;97:1714.

[52] Laing R. Image processing of corneal endothelial images. In: Ca-vanagh H (editor). The cornea: Transactions of the World Congresson the Cornea III. New York: Raven Press, 1988.

[53] Sherrard ER. Clinical specular microscopy of the corneal endothe-lium. Trans Ophthalmol Soc UK. 1981;101:156.

[54] Norn MS. Lissamine green. Vital staining of cornea and conjunctiva.Acta Ophthalmol (Copenhagen). 1973;51:483–491.

[55] Toda I, Tsubota K. Practical double vital staining for ocular surfaceevaluation. Cornea. 1993;12:366–367.

[56] Feenstra RPG, Tseng SCG. Comparison of fluorescein and rose ben-gal staining. Ophthalmology. 1992;110:984–993.

[57] Palmberg P. Gonioscopy. In: Ritch R, Shields MB, Krupin T (editors).The Glaucomas, Vol. I, 4th ed. St. Louis, MO: Mosby, 1996; 455–470.

[58] Kass MA. Standardizing the measurement of intraocular pressure forclinical research—Guidelines from the Eye Care Technology Forum.Ophthalmology. 1996;103:183–185.

[59] Kanski JJ. The Glaucomas. In: Kanski JJ, et al. (editors). ClinicalOphthalmology—A Systematic Approach, 4th ed. Oxford, Boston:Butterworth-Heinemann, 1999; 183–262.

[60] Chylack LT, Wolfe JK, Singer DM, et al. The Lens Opacities Classifi-cation System III. Arch Ophthalmol. 1993;111:831–836.

[61] Wilkinson CP, Ferris FL, Klein RE, et al. Proposed international clinicaldiabetic retinopathy and diabetic macular edema disease severityscales. Ophthalmology. 2003;110:1677–1682.

[62] Early Treatment of Diabetic Retinopathy Study Report No.1: Pho-tocoagulation for diabetic macular edema. Arch Ophthalmol.1985;103:796–806.

[63] Klein R, Klein BEK, Magli YL, et al. An alternative method of gradingdiabetic retinopathy. Ophthalmology. 1986;93:1183–1187.

[64] Touitou Y, Portaluppi F, Smolensky MH, et al. Ethical principles andstandards for the conduct of human and animal biological rhythmresearch. Chronobiol Int. 2004;21(1):161–170.

153 SN-Diabetic Retinopathy Study