Age-related macular degeneration in ethnically diverse australia: melbourne collaborative cohort study

Page 1

Ophthalmic Epidemiology, 2015; 22(2): 75–84! Informa Healthcare USA, Inc.

ISSN: 0928-6586 print / 1744-5086 online

DOI: 10.3109/09286586.2015.1010688

ORIGINAL ARTICLE

Age-Related Macular Degeneration in EthnicallyDiverse Australia: Melbourne Collaborative

Cohort Study

Liubov D. Robman1, Fakir M. A. Islam1,*, Elaine W. T. Chong1,Madeleine K. M. Adams1, Julie A. Simpson2, Khin Zaw Aung1, Galina A. Makeyeva1,

John L. Hopper2, Dallas R. English2,3, Graham G. Giles2,3, Paul N. Baird1, andRobyn H. Guymer1

1Centre for Eye Research Australia, Royal Victorian Eye and Ear Hospital, University of Melbourne, Melbourne,Australia, 2Centre for Molecular, Environmental, Genetic and Analytic Epidemiology, School of Public Health,University of Melbourne, Melbourne, Australia, and 3Cancer Epidemiology Centre, Cancer Council Victoria,

Melbourne, Australia

ABSTRACT

Purpose: To determine and compare the prevalence of age-related macular degeneration (AMD) in olderAustralians of Anglo-Celtic and Southern European origin.

Methods: A total of 21,132 participants of the Melbourne Collaborative Cohort Study, aged 47–86 years, wereassessed for AMD in 2003–2007 with non-mydriatic fundus photography. Of these, 14% were born in SouthernEurope (Greece, Italy or Malta), with the remaining 86% of Anglo-Celtic origin, born in Australia, the UnitedKingdom or New Zealand.

Results: Overall, 2694 participants (12.7%) had early stages of AMD, defined as either one or more drusen�125 mm (with or without pigmentary abnormalities) or one or more drusen 63–124mm with pigmentaryabnormalities in a 6000-mm diameter grading grid, in the absence of late AMD in either eye. A total of122 participants (0.6%) had late AMD, defined as either geographic atrophy or neovascular AMD. In logisticregression analysis, adjusted for age, sex, smoking, education and physical activity, Southern Europeanscompared to Anglo-Celts had a higher prevalence of the early stages of AMD (odds ratio, OR, 1.15, 95%confidence interval, CI, 1.00–1.34), and lower prevalence of late AMD (OR 0.36, 95% CI 0.17–0.78).

Conclusions: Australians of Southern European origin have a higher prevalence of the early stages of AMD andlower prevalence of late AMD compared to those of Anglo-Celtic origin. Although AMD prevalence in the olderage group(s) of Southern Europeans could be underestimated due to disparity in participation rates, it is likelythat both lifestyle and genetic factors play their parts in differential AMD prevalence in these ethnic groups.

Keywords: Age-related macular degeneration (AMD), ethnicity, prevalence

INTRODUCTION

Age-related macular degeneration (AMD) is the pri-mary cause of blindness in industrialized countries,globally being ranked third after cataract and

glaucoma.1 In Australia, it is responsible for nearlyhalf of all legal blindness2,3 and it is estimated that thecontinuing increase in life expectancy will lead to adoubling in the number of people with AMD by 2025,with a substantial impact on quality of life and

*Currently at the Faculty of Health, ART and Design, Swinburne University of Technology, Melbourne, Australia.Correspondence: Liubov D. Robman, Centre for Eye Research Australia, 32 Gisborne Street, East Melbourne, VIC 3002, Australia. Tel: +61 40578 0259. Fax: +61 3 9662 3859. E-mail: [email protected]

Received 23 March 2014; Revised 12 August 2014; Accepted 11 September 2014; Published online 17 March 2015

75

Oph

thal

mic

Epi

dem

iol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Mel

bour

ne o

n 05

/27/

15Fo

r pe

rson

al u

se o

nly.

Page 2

costs of care.4,5 Various studies of differentrigor, methodology, AMD definition and geographiclocation have reported prevalence of AMD for dif-ferent ethnicities, including studies of European des-cendants.6–19 The results from these studies indicateethnic disparities in AMD prevalence, with higherprevalence in Caucasians when compared to AfricanAmericans, Latinos, Asian Americans or Asian popu-lations.11,15–17,20–28 Even among different European-descendant populations there are differences inreported prevalence rates.6–19,29,30

A recent meta-analysis of AMD prevalence esti-mates in different populations of European ancestryshowed considerable differences, however therewere also differences in study characteristics, dis-ease definitions and age of participants acrossstudies, making it difficult to directly compareprevalence rates.19 With data being collected fromdifferent countries, other variables could also con-tribute to different ascertainment biases, making itmore difficult to draw conclusions about truedifferences in prevalence rates between differentethnic groups.

Australia, as an immigrant nation, offers a rareopportunity to evaluate the prevalence of AMD inpersons of different ethnic background living inthe same country. This eliminates some of thepotential variables of combining data from differentcountries such as standards of living and access tohealthcare. A massive influx of migrants fromSouthern Europe during the mid-1900s offered aunique opportunity to study morbidity in thisinteresting ethnic group, which has been reportedto have lower prevalence rates of other chronicdiseases of aging and greater longevity.31–36 To datethere are only limited data on the prevalence ofAMD in Southern Europeans and it is not clear fromthe available data whether the detected advanta-geous patterns are reflected also in the prevalence ofAMD.10,13,14,29,30

The Melbourne Collaborative Cohort Study(MCCS) is a single, large, ethnically diverse study,designed to investigate the role of diet and lifestyle inthe causation of common chronic diseases and pos-sible interactions between environmental exposuresand common genetic variants.37,38 This cohort wasdesigned specifically to include Australians of diverseethnicity, with one quarter of participants beingdrawn from a Southern European background, beingborn in Greece, Italy or Malta, and the remaindercoming from Anglo-Celtic origins, born in Australia,United Kingdom or New Zealand. To further under-stand the AMD risk in persons of different back-grounds in Australia, we evaluated and compared theprevalence of AMD in people of Anglo-Celticand Southern European backgrounds taking part inthe MCCS.

MATERIAL AND METHODS

Study Design

AMD prevalence was evaluated between 2003 and2007, at the follow-up assessment of the MCCS cohort.

Study Population

The MCCS is a volunteer-based prospective cohortstudy of 41,514 people, predominantly (99.3%) aged40–69 years at baseline (1990–1994), with approxi-mately equal proportions of participants across thesethree decades of age. To enhance the range of dietaryand genetic variation, the MCCS recruited partici-pants of Anglo-Celtic origin, born in Australia, the UKor New Zealand (75% of cohort), as well as SouthernEuropeans, born in Italy, Greece or Malta, who arrivedin Australia in the 1950s (25% of cohort). At baseline,clinical examinations were conducted and compre-hensive data on lifestyle, dietary intakes and healthconditions, as well as blood samples for DNA testing,were collected. Detection of AMD and other funduspathology was conducted concurrently with theMCCS follow-up examinations from 2003–2007,when participants were aged 48–86 years.

Data Collection and Ethics Approval

This study was a collaboration between the CancerCouncil Victoria, who recruited participants andcollected demographic, anthropometric, lifestyle andnutritional data, as well as blood samples, and theCentre for Eye Research Australia, who conductedfundus photography, grading of macular photos andvalidation of ophthalmic information. The study wasapproved by the Human Research & EthicsCommittees of The Cancer Council Victoria and theRoyal Victorian Eye and Ear Hospital. All participantssigned a written informed consent form after explan-ation of the study, consenting to the examinations,access to their medical records and also to passivefollow-up conducted through record linkage to elect-oral rolls, electronic telephone books, the VictorianCancer Registry, and death records. The studyadhered to the Declaration of Helsinki for researchon humans.

Detection of AMD

Detection of AMD and other fundus pathology wasconducted by using non-mydriatic 45�digital colorimages centered on the macula and optic disc, asdescribed previously.39,40 Images were graded using

76 L. D. Robman et al.

Ophthalmic Epidemiology

Oph

thal

mic

Epi

dem

iol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Mel

bour

ne o

n 05

/27/

15Fo

r pe

rson

al u

se o

nly.

Page 3

OptoLite/OptoMize Pro software (Digital HealthCareImage Management Systems, Cambridge, UK) accord-ing to the International Classification System for AMDgrading.39,41 All cases assigned to geographic atrophy(GA) or choroidal neovascular disease (CNV) and alsocases of other retinal pathology were reviewed andadjudicated (LR and RG). Images were graded with-out knowledge of baseline characteristics.

Participants were allocated to a single AMDcategory according to the status of the more affectedeye. Early stages of AMD (correspondent to inter-mediate AMD in the Beckman classification42) weredefined as the presence of one or more drusen�125 mm in size (with or without pigmentary abnorm-alities) or one or more drusen 63–124mm withpigmentary abnormalities in a 6000-mm diametergrading grid centered on the fovea, in the absence oflate AMD in either eye. Late AMD included GA,defined as a sharply delineated area of hypo- ordepigmentation �175 mm in diameter with morevisible choroidal vessels than in surrounding areas,or CNV, defined as retinal pigment epitheliumdetachment, subretinal neovascular membrane, cen-tral retinal scar, sub-retinal hemorrhages and/or hardexudates within the macular area, not related to otherretinal vascular disease. The reference group wasthose with no early stages of AMD or late AMD ineither eye; drusen 563 mm, pigmentary abnormalitiesonly, and also drusen 63–124mm without pigmentaryabnormalities were allowed in the reference group.There were no other retinal imaging modalitiesavailable on this cohort. We included in the analysisonly participants whose photos from both eyes wereof good enough quality for AMD grading.

Statistical Analysis

We report the baseline characteristics of participantsand non-participants in the follow-up study and alsoparticipant characteristics separately for each countryof birth. We estimated the prevalence of the earlystages of AMD and late AMD according to theircountry of birth or grouped country of birth (Anglo-Celts – Australia/New Zealand/UK, or SouthernEuropeans – Italy/Greece/Malta). Age standardiza-tion of prevalences was conducted with the totalAustralian data being the standard (i.e. directstandardization).43

All analyses were adjusted for age, sex andsmoking status (base model). The following potentialconfounders were added provisionally to the basemodel but were retained only if they changed theb-coefficients (loge odds ratios) by more than 5%:educational status, alcohol intake, red meat intake,saturated fat intake, total energy intake, serum chol-esterol level, glycemic load, vitamin C, vitamin E, b-

carotene, zinc, lutein/zeaxanthin, trans-unsaturatedfatty acids, omega-3 fatty acids, vegetable intake, fishintake, supplement use (vitamin C, vitamin E, codliver oil, or fish oil), waist-hip ratio, physical activityscore, and history of angina, heart attack, stroke ordiabetes. The final model of logistic regression ana-lysis determining the odds ratios (ORs) and 95%confidence intervals (CIs) for two primary outcomes(early stages of AMD and late AMD) in associationwith country of birth was adjusted for age, sex,smoking status (never smokers, former smokers orcurrent smokers), education (high school or higher)and physical activity score. The categorical variable ofage was used, as there was a non-linear associationbetween age and AMD. Effect modification by age,sex and smoking was assessed by fitting interactionterms between these variables and the groupedcountry of birth, and tested using likelihood ratiotests. Age-specific analyses were presented because ofstrong evidence of interaction between AMD and age.We limited the analysis for late AMD to the age group70 years or older, as there was an insufficient numberof late AMD cases in the younger age groups. Allanalyses were performed in SPSS version 19.0.1 (SPSSInc, Chicago, IL, USA).

RESULTS

Participation and BackgroundCharacteristics

Of the 41,514 MCCS participants enrolled at baselinein 1990–1994, 27,883 (67.2%) attended follow-upexaminations from May 2003 to June 2007. A flowchart of the changes in sample size from the baselinerecruitment to the sample analyzed for AMD ispresented in Figure 1. A total of 13,631 participantsdid not participate in MCCS follow-up due to death,illness, refusal, leaving Victoria or Australia, orunknown reasons. A further 5477 did not have eyeimages taken due to lack of photographic facilities atsome of the clinics. A total of 22,406 participants(54% of the original cohort) attended sessions forfundus photography. In 464, photos were missing orungradable for both eyes. We excluded from theanalysis 655 participants with no AMD features inone eye and missing or ungradable images from theother eye, as the presence of AMD in these partici-pants could not be determined. A total of 155participants with early stages of AMD in one eyeand missing or ungradable photos from the other eyewere also excluded from the analysis, as their lateAMD status in the fellow eye could not bedetermined. These records were retained for futureanalyses with the outcome ‘‘any AMD’’. A total of21,132 participants with complete data on AMD

AMD in Ethnically Diverse Australia: Melbourne Collaborative Cohort Study 77

! 2015 Informa Healthcare USA, Inc.

Oph

thal

mic

Epi

dem

iol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Mel

bour

ne o

n 05

/27/

15Fo

r pe

rson

al u

se o

nly.

Page 4

grading for both eyes were included in the analysis.The comparison of those available and not availablefor the analysis is presented in Table 1.

Participation rates of eligible individuals who werealive and living in Victoria at the time of therecruitment into the MCCS follow-up was similarfor those of Anglo-Celtic background; for those bornin Australia and New Zealand (20,667/25,365, 81.5%)and for those born in the UK (1886/2332, 80.9%).Participation rates were lower for those of SouthernEuropean background, born in Italy (3016/4810,62.7%), Greece (2072/4067, 50.9%) or Malta (242/367, 65.9%). Those who participated in the follow-upwere on average approximately 2 years younger,smoked less and had higher educational levels thanthose who did not participate.

We were not able to recruit 9.5% of Greek, 10.8% ofItalian, 8.1% of UK, 8.8% of Australian/New Zealandand 6.1% of Maltese participants because they had

died before or during the follow-up stage ofthe MCCS.

All migrant groups, on average, had lived inAustralia over 40 years, with the mean (standarddeviation, SD) number of years since immigration toAustralia for those who arrived from the UK,Italy, Greece or Malta being 42.2 (11.4) years, 47.3(6.7) years, 43.8 (6.4) years, and 46.5 (7.8) years,respectively.

Background characteristics are summarized inTable 1. The average age at baseline was similar forthe different ethnic groups. Women consistentlyparticipated at both baseline and follow-up moreoften than men in all ethnic groups, especially thoseborn in Australia/New Zealand. Those who had eversmoked were 8–12% more prevalent among the UK-born participants than in the other groups, whileprevalence of current smokers in the participants bornin every Southern European country was significantlyhigher than in those born in Australia/New Zealandor the UK (Table 1). Many of the participants fromItaly, Greece or Malta completed only primary school(56.7%, 61.4% and 28.2%, respectively), whereasalmost all Anglo-Celtic participants had high schoolor higher educational levels.

AMD Prevalence

AMD status was determined in 21,132 participants,of whom 314 (1.5%) had a combination of drusen63–124mm in size and retinal pigment epitheliumabnormalities in the same eye, and 2380 (11.3%) haddrusen �125 mm in size. Late AMD was present in 122participants; 67 (0.3%) had GA and 55 (0.26%) hadneovascular AMD in at least one eye. Prevalence ofAMD by country of birth is presented in Table 2.

Early Stages of AMDThe prevalence of the early stages of AMD was higherfor participants born in Italy (15.4%), Greece (15.3%)or Malta (15.3%) than for those born in Australia/New Zealand (12.5%) and the UK (10.4%); p50.001.After adjustment for age, sex, smoking status, educa-tional level and physical activity, the OR for prevalentearly stages of AMD for UK-born participants waslower compared to Australian-born participants,whereas it tended to be slightly higher for thoseborn in each of the Southern European countries(Table 3). When we combined Greek, Italian andMaltese participants into a Southern European groupand the UK and Australia/New Zealand-born into anAnglo-Celtic group, Southern Europeans had a higherprevalence of these early stages of AMD comparedto those of Anglo-Celtic origin (OR 1.15. 95% CI1.00–1.34).

We found evidence for effect modification by agegrouped in decades for both early stages of AMD and

Total MCCS sample (1990-1994)N=41,514

Born in: AUS/NZ 28,545UK 2,641Italy 5,411Greece 4,525Malta 392

Chose not topar�cipate

MCCS follow upor lost contact

9,058 (22%)

Par�cipated in MCCS follow-up(May 2003-June 2007)

27,883 (67%)

Photographed22,406

Not analysed due toungradeable ormissing photos

1,119 (2.7%)

Analysed for AMD21,287 (51%)

Not photographeddue to lack of photo

facili�es5,477 (13%)

Le� Victoria or Australia beforefollow-up examina�ons started

745 (1.8%)

Le� Victoria or Australia duringfollow-up stage, before being

examined74 (0.2%)

Died before follow upexamina�ons started

2,497 (6%)

Died during follow up stage,before being examined

1,257 (3%)

Born in: AUS/NZ 20,667UK 1,886Italy 3,016Greece 2,072Malta 242

Born in: AUS/NZ 16,898UK 1,492Italy 1,823GreeceMalta 165

909

FIGURE 1. Participation in the AMD sub-study of theMelbourne Collaborative Cohort Study (MCCS), Australia,1990–2007 (AMD, age-related macular degeneration; AUS/NZ, Australia/New Zealand).

78 L. D. Robman et al.

Ophthalmic Epidemiology

Oph

thal

mic

Epi

dem

iol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Mel

bour

ne o

n 05

/27/

15Fo

r pe

rson

al u

se o

nly.

Page 5

TA

BL

E1.

Bas

elin

ech

arac

teri

stic

so

fp

arti

cip

ants

incl

ud

edan

dn

ot

incl

ud

edin

AM

Dan

aly

sis:

the

Mel

bo

urn

eC

oll

abo

rati

ve

Co

ho

rtS

tud

y(2

003–

2007

),A

ust

rali

a.

Co

un

try

of

bir

thfo

rp

arti

cip

ants

incl

ud

edin

AM

Dan

aly

sis

Bas

elin

ech

arac

teri

stic

No

tin

clu

ded

N=

20,3

82In

clu

ded

N=

21,1

32T

-tes

to

r�

2

pv

alu

eA

US

/N

ZN

=16

,791

UK

N=

1478

Ital

yN

=18

04G

reec

eN

=89

6M

alta

N=

163

Ag

eat

bas

elin

e,m

ean

yea

rs(S

D)

57.0

(8.6

)53

.8(8

.5)

50.

001

53.8

(8.7

)54

.0(8

.3)

54.3

(7.7

)52

.8(7

.2)

51.0

(7.2

)A

ge

atfo

llo

w-u

p,

mea

ny

ears

(SD

)N

/A

65.2

(8.7

)65

.0(8

.8)

64.6

(8.4

)67

.6(8

.0)

66.3

(7.4

)62

.3(7

.3)

Par

tici

pan

tsin

age

gro

up

sat

foll

ow

up

,n

(%)

560

yea

rsN

/A

7071

(33.

5)58

55(3

4.9)

514

(34.

8)41

9(2

3.2)

210

(23.

4)73

(44.

8)60

–70

yea

rsN

/A

7002

(33.

1)53

91(3

2.1)

525

(35.

5)63

1(3

5.0)

389

(43.

4)66

(40.

5)70

–80

yea

rsN

/A

6350

(30.

0)50

07(2

9.8)

408

(27.

6)64

5(3

5.8)

269

(30.

0)21

(12.

9)80

+y

ears

N/

A70

9(3

.4)

538

(3.2

)31

(2.1

)10

9(6

.0)

28(3

.1)

3(1

.8)

Mal

e,n

(%)

8618

(42.

3)84

27(3

9.9)

50.

001

6461

(38.

5)67

3(4

5.5)

805

(44.

6)41

6(4

6.4)

72(4

4.2)

Sm

ok

ing

stat

us,

n(%

)N

ever

smo

ked

11,0

50(5

4.2)

12,7

69(6

0.4)

50.

001

10,2

99(6

1.3)

754

(51.

0)10

61(5

8.9)

568

(63.

4)87

(53.

4)F

orm

ersm

ok

er64

30(3

1.6)

6567

(31.

1)51

57(3

0.7)

594

(40.

2)54

6(3

0.3)

222

(24.

8)48

(29.

4)C

urr

ent

smo

ker

2894

(14.

2)17

94(8

.5)

1335

(8.0

)13

0(8

.8)

195

(10.

8)10

6(1

1.8)

28(1

7.2)

Hig

hsc

ho

ol

or

hig

her

edu

cati

on

,n

(%)

14,4

08(7

0.7)

19,0

51(9

0.2)

50.

001

16,3

62(9

7.4)

1445

(97.

8)78

1(4

3.3)

346

(38.

6)11

7(7

1.8)

Alc

oh

ol

inta

ke,

med

ian

g/

day

(IQ

R)

2.1

(15.

8)5.

5(1

7.6)

50.

001

5.5

(17.

4)8.

9(2

0.4)

7.2

(23.

2)0.

2(9

.6)

2.0

(8.9

)R

edm

eat

inta

ke,

med

ian

tim

es/

wee

k(I

QR

)7.

0(5

.0)

7.0

(5.0

)0.

327.

0(4

.5)

6.5

(4.5

)7.

5(5

.5)

8.0

(6.5

)8.

5(7

.5)

Sat

ura

ted

fat

inta

ke,

med

ian

g/

day

(IQ

R)

75.1

(40.

8)78

.7(3

9.6)

50.

001

79.7

(39.

6)77

.9(3

7.5)

68.0

(36.

5)82

.1(4

4.4)

87.2

(45.

8)S

eru

mch

ole

ster

ol,

med

ian

mm

ol/

L(I

QR

)5.

6(1

.42)

5.4

(1.3

)5

0.00

15.

3(1

.3)

5.4

(1.3

)5.

4(1

.3)

5.4

(1.3

)5.

5(1

.3)

To

tal

ener

gy

inta

ke,

med

ian

kj/

day

(IQ

R)

8475

(415

6)89

13(3

936)

50.

001

9004

(391

9)90

50(3

852)

8003

(369

5)85

68(4

265)

9249

(528

7)G

lyce

mic

load

,m

edia

n(I

QR

)11

0.2

(64.

9)11

7.1

(62.

7)5

0.00

111

8.7

(62.

4)11

8.3

(60.

6)10

6.5

(62.

3)10

6.0

(61.

3)13

6.4

(81.

6)P

hy

sica

lac

tiv

ity

sco

re,

mea

n(S

D)

3.62

(3.3

9)4.

38(3

.69)

50.

001

4.59

(3.7

2)5.

05(3

.89)

2.72

(2.8

6)2.

71(2

.88)

3.04

(2.9

5)W

aist

-hip

rati

o,

mea

n(S

D)

0.85

(0.0

9)0.

83(0

.09)

50.

001

0.82

(0.0

9)0.

83(0

.09)

0.86

(0.0

9)0.

86(0

.09)

0.82

(0.0

9)H

isto

ryo

fan

gin

a,n

(%)

1058

(5.2

)53

6(2

.5)

50.

001

450

(2.7

)27

(1.8

)36

(2.0

)18

(2.0

)5

(3.1

)H

isto

ryo

fh

eart

atta

ck,

n(%

)76

0(3

.7)

403

(1.9

)5

0.00

133

4(2

.0)

20(1

.4)

30(1

.7)

16(1

.8)

3(1

.8)

His

tory

of

stro

ke,

n(%

)35

7(1

.8)

151

(0.7

)5

0.00

111

1(0

.7)

12(0

.8)

25(1

.4)

3(0

.3)

0(0

)H

isto

ryo

fd

iab

etes

,n

(%)

1175

(5.8

)37

3(1

.8)

50.

001

217

(1.3

)25

(1.7

)80

(4.4

)45

(5.0

)6

(3.7

)

AM

D,

age-

rela

ted

mac

ula

rd

egen

erat

ion

;A

US

/N

Z,

Au

stra

lia/

New

Zea

lan

d;

IQR

,in

terq

uar

tile

ran

ge;

SD

,st

and

ard

dev

iati

on

.

AMD in Ethnically Diverse Australia: Melbourne Collaborative Cohort Study 79

! 2015 Informa Healthcare USA, Inc.

Oph

thal

mic

Epi

dem

iol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Mel

bour

ne o

n 05

/27/

15Fo

r pe

rson

al u

se o

nly.

Page 6

late AMD (log likelihood ratio test p50.008 andp50.001, respectively), therefore for the early stagesof AMD we presented the comparative analysisstratified by age groups, whereas for late AMD thecomparison is presented for the age group of those70 years or older, as there were not enough cases forstatistical analysis in the younger age groups. The lownumber of Maltese participants in the older agegroups did not allow us to obtain estimates for lateAMD and also for the early stages of AMD in theoldest age group (80 years or older). The higher ORsfor the early stages of AMD in Southern Europeanscompared to those of Anglo-Celtic backgroundappeared predominantly in the age group 60-69years, whereas in the older age groups it was notdifferent and tended to be lower in Southern

Europeans, although not statistically significant(Table 3, Figure 2).

Late AMDIn total, there were 122/21,132 (0.6%) cases of lateAMD in this study. There was no difference in theprevalence of GA compared to CNV in the totalsample (data not shown). The prevalence of late AMDwas 0.6%, 0.4%, 0.6%, 0.3%, and 0.6% among theAustralian/New Zealand, UK, Italian, Greek, andMaltese groups, respectively (Table 2). For late AMD,after adjustment for covariates, Southern Europeanshad lower prevalence of late AMD compared tothose of Anglo-Celtic origin (OR 0.36, 95% CI 0.17–0.78; Table 3).

TABLE 3. Associations of AMD OR (95% CI)a with country of birth or grouped country of birth in the Melbourne CollaborativeCohort Study (2003–2007), Australia.

AUS/NZ UK Italy Greece Malta Anglo-CelticSouthernEuropean

Early stages of AMDb

Total sample 1.0 (ref) 0.81 (0.68–0.97) 1.11 (0.93–1.31) 1.16 (0.93–1.43) 1.38 (0.90–2.14) 1.0 (ref) 1.15 (1.00–1.34)Age, years

Under 60 1.0 (ref) 0.60 (0.42–0.87) 0.97 (0.69–1.37) 1.07 (0.66–1.71) 2.15 (1.18–3.92) 1.0 (ref) 1.15 (0.87–1.53)60–69 1.0 (ref) 0.86 (0.63–1.17) 1.82 (1.36–2.43) 1.45 (1.01–2.08) 1.05 (0.47–2.32) 1.0 (ref) 1.64 (1.27–2.12)70–79 1.0 (ref) 0.85 (0.64–1.13) 0.87 (0.66–1.15) 1.12 (0.79–1.60) 1.13 (0.37–3.39) 1.0 (ref) 0.96 (0.75–1.23)80 or older 1.0 (ref) 2.23 (1.02–4.90) 0.75 (0.39–1.44) 0.60 (0.21–1.76) N/A 1.0 (ref) 0.67 (0.36–1.23)

Late AMDc 1.0 (ref) 0.86 (0.37–1.99) 0.41 (0.18–0.93) 0.25 (0.06–1.11) N/A 1.0 (ref) 0.36 (0.17–0.78)

aORs (95% CIs) adjusted for age (5-year categories for total sample and continuous scale in groups stratified by decades of age), sex,smoking status (never, former and current), education (below high school vs. high school or above education) and physical activityscore.bEarly stages of AMD defined as drusen 63–124 mm with pigmentary abnormalities or drusen �125 mm with or without pigmentaryabnormalities.cLate AMD defined as either geographic atrophy or neovascular AMD.Bolding denotes significant values.AMD, age-related macular degeneration; AUS/NZ, Australia/New Zealand; CI, confidence interval; OR, odds ratio; ref, reference.

TABLE 2. Prevalence of AMD by country of birth in the Melbourne Collaborative Cohort Study (2003–2007), Australia.

No AMD Early stages of AMDa Late AMDb

n

Observedprevalence,

%

Age-standardizedprevalence,

% n

Observedprevalence,

%

Age-standardizedprevalence,

% n

Observedprevalence,

%

Age-standardizedprevalence,

%

Total sample 18,316 86.7 95.1 2694 12.7 13.9 122 0.6 1.0AUS/NZ 14,589 86.9 85.4 2101 12.5 13.5 101 0.6 1.1UK 1319 89.2 86.1 153 10.4 13.5 6 0.4 0.4Italy 1515 84.0 83.6 278 15.4 15.3 11 0.6 1.1Greece 756 84.4 84.0 137 15.3 15.2 3 0.3 0.8Malta 137 84.1 85.6 25 15.3 13.9 1 0.6 0.5Anglo-Celts 15,908 87.1 85.4 2254 12.3 13.5 107 0.6 1.1Southern

Europeans2408 84.1 83.6 440 15.4 15.4 15 0.5 1.0

aEarly stages of AMD defined as drusen 63–124 mm with pigmentary abnormalities or drusen �125 mm with or without pigmentaryabnormalities.bLate AMD defined as either geographic atrophy or neovascular AMD.AMD, age-related macular degeneration; AUS/NZ, Australia/New Zealand.

80 L. D. Robman et al.

Ophthalmic Epidemiology

Oph

thal

mic

Epi

dem

iol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Mel

bour

ne o

n 05

/27/

15Fo

r pe

rson

al u

se o

nly.

Page 7

DISCUSSION

In this large cohort of volunteers examined for AMDin Melbourne, overall prevalence rates for the earlystages of AMD (12.7%) and late AMD (0.6%) weresimilar to those reported in the multi-ethnicMelbourne Visual Impairment Project (MelbourneVIP; 1992–1994), undertaken in the same community,including a proportion of participants of Greek andItalian origin.12 For drusen �125 mm, both the MCCSand Melbourne VIP reported 11% prevalence, eventhough different methods of fundus photography(digital non-mydriatic versus film-based dilated,respectively) and different study designs (volunteer-based versus population-based, respectively) wereemployed.12 Another Australian population, the BlueMountain Eye Study (BMES), which enrolled pre-dominantly European descendants of Anglo-Celticorigin, reported a slightly higher prevalence of drusen�125 mm (13.3%) and higher prevalence of late AMD(1.9%)9. Given our finding of a lower prevalence rateof late AMD in the Greek and Italian sub-populations,it is possible that the large Greek and Italian presencein the Melbourne population contributed to the lowerprevalence of late AMD in the VIP compared to theBMES.

The limited data on AMD prevalence in Greek andItalian populations varies in methods of collection anddefinitions.10,13,29,30 One comparative study of differ-ent ethnic groups is the international multicenterEuropean Eye (EUREYE) study,14 where 4753 partici-pants aged 65 years or older were enrolled by 2002from seven countries across Europe. This studyincluded 605 Italians and 587 Greeks, which isconsiderably smaller than the MCCS sub-popula-tions of 1804 Italian and 896 Greek participants.The EUREYE study reported age- and sex-standar-dized prevalence of AMD by study center usingthe total population as the standard. While

statistical methods differ between our study and theEUREYE study (direct comparison versus standard-ization), and different covariates were used foradjustment in the analyses, it is of interest to comparethe results.

As the age of enrolled participants in MCCS wasyounger than in EUREYE, we compared the preva-lence of drusen �125 mm or late AMD in those 10,505participants aged �65 years only from the MCCS (thelower age limit at the EUREYE study). The overallprevalence of drusen �125 mm in EUREYE comparedwith MCCS was similar, 15.4% versus 13.8%, respect-ively, but the overall prevalence of late AMD washigher: 3.32% in the EUREYE compared to 1.1% in theMCCS. In the Italian and Greek sub-samples of theEUREYE, large drusen (�125 mm) had prevalences of13.7% and 14.8%, respectively, which was slightlylower than the 15.4% in the Italian and 15.4% in theGreek MCCS participants aged 65 year or older.However, the prevalences of late AMD were muchhigher in the EUREYE participants (3.7% in Italiansand 4.7% in Greeks) compared to the MCCS partici-pants born in Italy (0.9%) or Greece (0.4%). TheEUREYE study also detected late AMD prevalence of3.77% in the UK cohort from Northern Ireland(Belfast), which again was higher than the late AMDprevalence of 0.9% in the UK-born MCCS participantsaged 65 years or older. It is possible that the estimatesfor the lower total prevalence of late AMD in theMCCS could be partially explained by the upper agelimit in the MCCS being 86 years, with 709 (3.4% ofthe study sample) being 80–86 years old (mean age inthis group was 81 ± 1.1 years), while it is not clearwhat upper age was present in the EUREYE cohortnor the proportions of those aged 80–90 years andover 90 years. However it is also possible that thedifferences in prevalence of late AMD could be a truefinding suggesting that those living in Australia,irrespective of ethnicity, could be in some wayprotected from developing late AMD.

0

10

20

30

40

50

60

UK AUS/NZ Italy Greece Malta

Prev

alen

ce o

f ear

ly A

MD

stag

es (%

)

Country of birth

Under 60 60-69 70-79 80 or older

FIGURE 2. Prevalence of early stages of AMD in participants of the Melbourne Collaborative Cohort Study (2003–2007, N = 21,132) byage group (years) and country of birth (AMD, age-related macular degeneration; AUS/NZ, Australia/New Zealand).

AMD in Ethnically Diverse Australia: Melbourne Collaborative Cohort Study 81

! 2015 Informa Healthcare USA, Inc.

Oph

thal

mic

Epi

dem

iol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Mel

bour

ne o

n 05

/27/

15Fo

r pe

rson

al u

se o

nly.

Page 8

We found the OR for AMD differed by ethnicgroup, with those of Southern European origin havinghigher odds for the early stages of AMD, but lowerodds for late AMD, compared to those of Anglo-Celticorigins. We also found that the association of the earlystages of AMD with ethnicity was modified by age,with higher odds for early stage AMD in SouthernEuropeans compared to Anglo-Celts at the youngerage group.

The strengths of our study include its single largecohort, inclusive of a large sub-cohort of ethnic Greeksand Italians. Having one large cohort of co-locatedparticipants allowed for uniformity in study proced-ures across all participants, thus setting the stage forcomparison between ethnic groups. Despite being thelargest single cohort to investigate differences inprevalence, the numbers with late AMD remainsmall, so it is important not to over interpret the lateAMD results. A weakness of the study is that thediagnosis of AMD was made on the basis of only colorfundus images and so there is a possibility that somecases could have been misclassified. However we donot believe that the likelihood of this occurring wasmore likely in one ethnic group over another.

The reversal of the direction of association betweenAMD and Southern European origin from positive forearly stages of AMD to negative for late AMD isinteresting, especially when considering the higherprevalence of current smokers in the SouthernEuropean group. The combination of a well-recog-nized direct association between smoking and AMDon the one hand,44 the inverse association betweensmoking and survivorship on the other hand,45 andalso the dependence of AMD prevalence on longevitymakes statistical analysis and interpretation morecomplex, where survival and participation bias couldinfluence the results to a greater degree than fordiseases that manifest at younger ages than AMD.

In the MCCS, as with any elderly cohort, there is thepotential for a stronger effect of survivorship on theassociations in question, with only those remainingalive being at risk of AMD and only those healthyenough to attend the study being able to participate.The phenomenon of the ‘‘healthy migrant effect’’ hadthe potential to bias towards lower morbidity andmortality rates in the ethnic Greeks and Italians.However, early 1990s data showed higher rates ofmorbidity and cardiovascular risk factors (obesity,diabetes, hyperlipidemia, hypertension) for Greek orItalian middle-aged migrants, indicating that by theearly 1990s the healthy migrant effect did not appear topersist.46–51 Ethnically determined differential effectsof poorer health, lowering mobility and increasingmorbidity that reduce the likelihood of participation,could contribute to unequal participation rates anddifferential survivorship effect. However, the observedhigher prevalence of early AMD, yet lower prevalenceof late AMD in the Southern European group

compared to the Anglo-Celtic group could reflect atrue difference and would suggest that different factorsinfluence the various stages of the disease.

The most striking findings from our study are thateven within a community that have lived together forover 40 years there appear to be differences in diseaseprevalence based on European ethnicity. Results fromthis very large cohort suggest that ethnic influences,likely both lifestyle and genetic factors, play theirparts in determining the prevalence of AMD. Thereported differences in prevalence rates of AMDamong similar ethnicities living in different parts ofthe world provide some suggestion that environmen-tal influences are particularly important in under-standing the differences found.

ACKNOWLEDGMENTS

GGG, DRE and JLH designed, raised funds andconducted the parent MCCS; LDR, RHG, and PNBdesigned and raised funds for AMD research in theMCCS. EWC, KZA, GAM, LDR, MKMA and RHGcollected and validated the ophthalmic data. LDR andFMAI conducted the statistical analysis, MKMA, RHGand JAS contributed to data analysis. LDR and RHGsupervised the ophthalmic part of the MCCS andwrote the paper.

All authors critically revised the manuscript,contributed to writing and approved the final versionfor submission.

DECLARATION OF INTEREST

The authors report no conflicts of interest. The authorsalone are responsible for the content and writing ofthe paper.

Cohort recruitment was funded by VicHealth andThe Cancer Council Victoria. Further MCCS funding:the National Health & Medical Research Council ofAustralia (NHMRC) Program Grant 209057, CapacityBuilding Grant 251533 and Enabling Grant 396414. Theophthalmic component was funded by the OphthalmicResearch Institute of Australia; American HealthAssistance Foundation, Jack Brockhoff Foundation,John Reid Charitable Trust, Perpetual Trustees andRoyal Victorian Eye and Ear Hospital. People supportwas provided through the NHMRC CareerDevelopment (300052) and Practitioner (529905)Fellowships to RHG, Wagstaff Fellowship to LR,NHMRC PhD scholarship (590226) and Hugh NoelPuckle Scholarship to MA, NHMRC Senior ResearchFellowships to PNB (1028444) and JH (1023434), andNovartis Medical Retinal Fellowship to EC. Centre forEye Research Australia is a recipient of the NHMRCCentre for Clinical Research Excellence Grant 529923and Operational Infrastructure Support from the

82 L. D. Robman et al.

Ophthalmic Epidemiology

Oph

thal

mic

Epi

dem

iol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Mel

bour

ne o

n 05

/27/

15Fo

r pe

rson

al u

se o

nly.

Page 9

Victorian Government. The funding organizations didnot have any involvement in study design and datacollection, analysis or interpretation.

REFERENCES

1. WHO. Causes of blindness and visual impairment. Priorityeye diseases: Age-related macular degeneration 2013.Available from: http://www.who.int/blindness/causes/priority/en/index8.html [last accessed 23 Dec 2013].

2. Attebo K, Mitchell P, Smith W. Visual acuity and the causesof visual loss in Australia. The Blue Mountains Eye Study.Ophthalmology 1996;103:357–364.

3. Weih LM, VanNewkirk MR, McCarty CA, et al. Age-specific causes of bilateral visual impairment. ArchOphthalmol 2000;118:264–269.

4. Access Economics. Centrally focussed ‘‘The impact of Age-related Macular Degeneration’’ – A dynamic economic modeland report. Melbourne: Eye Research Australia, 2006.

5. Deloitte Access Economics. Eyes on the future. A clearoutlook on Age-related Macular Degeneration. MacularDegeneration Foundation. 2011:145 pp.

6. Gibson J, Rosenthal A, Lavery J. A study on the prevalenceof eye disease in the elderly in an English community.Trans Ophthalmol Soc UK 1985;104:196–203.

7. Klein R, Klein BE, Linton KL. Prevalence of age-relatedmaculopathy. The Beaver Dam Eye Study. Ophthalmology1992;99:933–943.

8. Vingerling JR, Dielemans I, Hofman A, et al. The preva-lence of age-related maculopathy in the Rotterdam Study.Ophthalmology 1995;102:205–210.

9. Mitchell P, Smith W, Attebo K, et al. Prevalence of age-related maculopathy in Australia. The Blue Mountains EyeStudy. Ophthalmology 1995;102:1450–1460.

10. Pagliarini S, Moramarco A, Wormald RP, et al. Age-relatedmacular disease in rural southern Italy. Arch Ophthalmol1997;115:616–622.

11. Klein R, Klein B, Cruickshanks KJ. The prevalence of age-related maculopathy by geographic region and ethnicity.Prog Ret Eye Res 1999;18:371–389.

12. VanNewkirk MR, Nanjan MB, Wang JJ, et al. The preva-lence of age-related maculopathy: the visual impairmentproject. Ophthalmology 2000;107:1593–1600.

13. Topouzis F, Coleman AL, Harris A, et al. Prevalence of age-related macular degeneration in Greece: the ThessalonikiEye Study. Am J Ophthalmol 2006;142:1076–1079.

14. Augood CA, Vingerling JR, de Jong PT, et al. Prevalence ofage-related maculopathy in older Europeans: the EuropeanEye Study (EUREYE). Arch Ophthalmol 2006;124:529–535.

15. Klein R, Klein BE, Knudtson MD, et al. Prevalence of age-related macular degeneration in 4 racial/ethnic groups inthe multi-ethnic study of atherosclerosis. Ophthalmology2006;113:373–380.

16. Klein R, Chou CF, Klein BE, et al. Prevalence of age-relatedmacular degeneration in the US population. ArchOphthalmol 2011;129:75–80.

17. Owen CG, Jarrar Z, Wormald R, et al. The estimatedprevalence and incidence of late stage age related maculardegeneration in the UK. BJO 2012;96:752–756.

18. Erke MG, Bertelsen G, Peto T, et al. Prevalence of age-related macular degeneration in elderly Caucasians: theTromso Eye Study. Ophthalmology 2012;119:1737–1743.

19. Rudnicka AR, Jarrar Z, Wormald R, et al. Age and gendervariations in age-related macular degeneration prevalencein populations of European ancestry: a meta-analysis.Ophthalmology 2012;119:571–580.

20. Friedman DS, Katz J, Bressler NM, et al. Racial differencesin the prevalence of age-related macular degeneration:the Baltimore Eye Survey. Ophthalmology 1999;106:1049–1055.

21. Friedman DS, O’Colmain BJ, Munoz B, et al. Prevalence ofage-related macular degeneration in the United States.Arch Ophthalmol 2004;122:564–572.

22. Li Y, Xu L, Jonas JB, et al. Prevalence of age-relatedmaculopathy in the adult population in China: the Beijingeye study. Am J Ophthalmol 2006;142:788–793.

23. Kawasaki R, Yasuda M, Song SJ, et al. The prevalence ofage-related macular degeneration in Asians: a systematicreview and meta-analysis. Ophthalmology 2010;117:921–927.

24. Nonyane BA, Nitsch D, Whittaker JC, et al. An ecologicalcorrelation study of late age-related macular degenerationand the complement factor H Y402H polymorphism. InvestOphthalmol Vis Sci 2010;51:2393–2402.

25. Krishnan T, Ravindran RD, Murthy GV, et al. Prevalence ofearly and late age-related macular degeneration in India:the INDEYE study. Invest Ophthalmol Vis Sci 2010;51:701–707.

26. Vanderbeek BL, Zacks DN, Talwar N, et al. Racial differ-ences in age-related macular degeneration rates in theUnited States: a longitudinal analysis of a managed carenetwork. Am J Ophthalmol 2011;152:273–282 e3.

27. Stein JD, Vanderbeek BL, Talwar N, et al. Rates ofnonexudative and exudative age-related macular degener-ation among Asian American ethnic groups. InvestOphthalmol Vis Sci 2011;52:6842–6848.

28. Yang K, Liang YB, Gao LQ, et al. Prevalence of age-related macular degeneration in a rural Chinese popula-tion: the Handan Eye Study. Ophthalmology 2011;118:1395–1401.

29. Carresi C, Cruciani F, Paolucci F, et al. Montelparo study:risk factors for age-related macular degeneration in a littlerural community in Italy. Clin Ter 2009;160:e43–51.

30. Spanish Eyes Epidemiological (SEE) Study Group.Prevalence of age-related macular degeneration in Spain.BJO 2011;95:931–936.

31. Trichopoulou A, Lagiou P, Trichopoulos D. TraditionalGreek diet and coronary heart disease. J Cardiovasc Risk1994;1:9–15.

32. Trichopoulou A, Vasilopoulou E. Mediterranean diet andlongevity. Br J Nutr 2000;84(Suppl. 2):S205–209.

33. Trichopoulou A, Costacou T, Bamia C, et al. Adherence to aMediterranean diet and survival in a Greek population.NEJM 2003;348:2599–2608.

34. Wang Y, Simpson JA, Wluka AE, et al. Reduced rates ofprimary joint replacement for osteoarthritis in Italian andGreek migrants to Australia: the Melbourne CollaborativeCohort Study. Arthritis Res Ther 2009;11:R86.

35. Vasto S, Scapagnini G, Rizzo C, et al. Mediterranean dietand longevity in Sicily: survey in a Sicani Mountainspopulation. Rejuvenation Res 2012;15:184–188.

36. WHO. World Health Statistics 2012, Part III Global HealthIndicators. Available from: http://www.who.int/healthinfo/EN_WHS2012_Part3.pdf [last accessed 19 Feb2015].

37. Giles G. The Melbourne Study of Diet and Cancer. ProcNutr Soc Aust 1990;15:61–68.

38. Giles G, English D. The Melbourne Collaborative CohortStudy. IARC – International Agency for Research onCancer. Lyon, France. 2002;156:69–70.

39. Aung KZ, Robman L, Chong EW, et al. Non-mydriaticdigital macular photography: how good is the second eyephotograph? Ophthalmic Epidemiol 2009;16:254–261.

40. Adams MK, Simpson JA, Aung KZ, et al. Abdominalobesity and age-related macular degeneration. Am JEpidemiol 2011;173:1246–1255.

AMD in Ethnically Diverse Australia: Melbourne Collaborative Cohort Study 83

! 2015 Informa Healthcare USA, Inc.

Oph

thal

mic

Epi

dem

iol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Mel

bour

ne o

n 05

/27/

15Fo

r pe

rson

al u

se o

nly.

Page 10

41. Bird AC, Bressler NM, Bressler SB, et al. An internationalclassification and grading system for age-related maculo-pathy and age-related macular degeneration. TheInternational ARM Epidemiological Study Group. SurvOphthalmol 1995;39:367–374.

42. Ferris III FL, Wilkinson CP, Bird A, et al. Clinical classi-fication of age-related macular degeneration.Ophthalmology 2013;120:844–851.

43. Australian Bureau of Statistics. 3235.0 – Population by Ageand Sex, Regions of Australia. 2012. Available from:http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/3235.02012?OpenDocument [last accessed 20 Jun 2014].

44. Smith W, Assink J, Klein R, et al. Risk factors for age-related macular degeneration: pooled findings from threecontinents. Ophthalmology 2001;108:697–704.

45. Shaw BA, Agahi N. A prospective cohort study of healthbehavior profiles after age 50 and mortality risk. BMCPublic Health 2012;12:803.

46. Ireland PD, Giles GG. Blood pressure in the Australian-and Italian-born: a Melbourne suburb revisited after20 years. Aust NZJ Public Health 1996;20:143–148.

47. Kouris-Blazos A, Wahlqvist ML, Trichopoulou A, et al.Health and nutritional status of elderly Greek migrants toMelbourne, Australia. Age Ageing 1996;25:177–189.

48. Kouris-Blazos A. Morbidity mortality paradox of 1stgeneration Greek Australians. Asia Pacific J Clin Nutr2002;11(Suppl. 3):S569–575.

49. Hodge AM, English DR, O’Dea K, et al. Increased diabetesincidence in Greek and Italian migrants to Australia: howmuch can be explained by known risk factors? DiabetesCare 2004;27:2330–2334.

50. Bennett SA. Inequalities in risk factors and cardiovascularmortality among Australia’s immigrants. Aust J PublicHealth 1993;17:251–261.

51. Kalantar-Zadeh K. What is so bad about reverse epidemi-ology anyway? Semin Dial 2007;20:593–601.

84 L. D. Robman et al.

Ophthalmic Epidemiology

Oph

thal

mic

Epi

dem

iol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Mel

bour

ne o

n 05

/27/

15Fo

r pe

rson

al u

se o

nly.