The Barbados Eye Study

Prevalence of Open Angle Glaucoma

M. Cristina Leske, MD, MPH; A. M. S. Connell, FRCS, FRCOphth; Andrew P. Schachat, MD;
Leslie Hyman, PhD; the Barbados Eye Study Group

Objective: To describe the design of the Barbados Eye were tested at the study site, 95% completed Humphrey
Study and report on the prevalence of open angle glau- automated perimetry and 97% had photographic or clini-
coma (OAG) in a predominantly black study popula- cal disc gradings; 93% of those referred completed the
tion. ophthalmologic examination. In this adult population,
the prevalence of OAG by self-reported race was 7.0%
Design: Population-based prevalence study. (302/4314) in black, 3.3% (6/184) in mixed-race, and 0.8%
(1/133) in white or other participants. In black and mixed-
Setting and Participants: Residents of Barbados, West race participants, the prevalence reached 12% at age 60

Indies, identified from a simple random sample of Bar- years and older and was higher in men (8.3%) than in
badian-born citizens 40 through 84 years old. women (5.7%), with an age-adjusted male-female ratio
of 1.4. In addition, over 3% of the participants were clas-
Data Collection: Participants had a comprehensive study sified as having suspect OAG.
visit that included automated perimetry, applanation to-
nometry, and fundus photography; persons with spe- Conclusions: To our knowledge, the Barbados Eye Study
cific examination findings, as well as a 10% sample of par- is the largest glaucoma study ever conducted in a black
ticipants, were referred for an ophthalmologic examination population and identified more people with OAG than
and additional tests. did any previous population study. The prevalence of OAG
was high, especially at older ages and in men. Among par-
Outcome: A diagnosis of OAG required both visual field ticipants 50 years old or older, one in 11 had OAG, and
and optic disc criteria for glaucoma damage after exclud- prevalence increased to one in six at age 70 years or older.
ing other causes. The results highlight the public health importance of OAG
in the Afro-Caribbean region and have implications for
Results: The 4709 participants (83.5% of those eli- other populations.
gible) had demographic characteristics that were similar
to the census population. Of the 4631 participants who (Arch Ophthalmol. 1994;112:821-829)

OPEN
ANGLE glaucoma miologic data on OAG, age-related cata¬
(OAG) is major
a cause ract, age-related macular degeneration, and
of visual impairment diabetic retinopathy in a predominantly
and blindness.1"3 A black population. To our knowledge, it was
higher frequency of the first comprehensive study of preva¬
OAG visual loss exists in black than in lence and risk factors for all the major eye
white populations, but the reasons for these diseases based on a simple random sample
From the School of Medicine,
differences are not well understood. I4° De¬ of a country's population; a major goal was
the State University of New
York-Stony Brook (Drs Leske spite its importance and public health rel¬
and Hyman), the Ministry of evance, limited information is available on
Health, Bridgetown, Barbados, the prevalence of OAG or other eye dis¬
West Indies (Dr Connell), and eases in black people and even less is
The Wilmer Ophthalmological See Methods next
Institute, The Johns
known about possible risk factors. on page
Hopkins
University School of Medicine, The Barbados Eye Study (BES) was
Baltimore, Md (Dr Schachat). designed to address the need for epide-

Downloaded From: http://archopht.jamanetwork.com/ by a GLASGOW UNIVERSITY LIBRARY User on 08/01/2013

 METHODS (waist and hips); iris color gradings (black, brown,
medium, and light); complexion gradings (very
dark, dark, medium, and light); and three applana-
The BES was funded by the National Eye Institute tion tonometry measurements in each eye. After a
and included a coordinating center (the State Uni¬ slit-lamp assessment of anterior chamber depth, pu¬
versity of New York-Stony Brook), a data collection pils were dilated with tropicamide. A standardized
center (the Ministry of Health, Bridgetown, Barba¬ interview on demographic, medical, family, environ¬
dos, West Indies) and a fundus photograph reading mental, and other risk factors was completed during
center (The Johns Hopkins University, Baltimore, dilatation. Lenses were graded at the slit lamp fol¬
Md). The study began in October 1987 with a lowing the Lens Opacities Classification System II
start-up and pretesting phase to train and standard¬ protocol.9 Bilateral external and 30° nonsimulta-
ize staff and to finalize recruitment methods and neous stereo color fundus photographs of the disc
data collection protocols. A total of 175 volunteers and macula (Early Treatment Diabetic Retinopathy
and patients with glaucoma participated in this Study standard fields 1 and 210) were taken, and film
phase. Data collection for the BES began in April was sent to the reading center for development and
1988 and ended by May 1992. grading. A blood sample was drawn to measure gly-
Persons eligible for the BES were current Barba¬ cosylated hemoglobin by a column method" (for
dos residents who were identified through a simple diabetes screening) and to determine ABO, Rh, and
random sample of Barbadian-born citizens 40 Duffy blood groups (to assess genetic markers). All
through 84 years old, which was selected by the Bar¬ data were collected by study personnel following
bados Statistical Services Department, Bridgetown. standard protocols and procedures; reproducibility
The sample was based on national registration num¬ of measurements and gradings was monitored
bers, which provide a required and unique identifier throughout the study.
for every Barbadian by incorporating date of birth, All persons with an intraocular pressure (IOP)
sex, and birthplace.6 A sample of 7600 national reg¬ over 21 mm Hg in either eye, abnormal visual fields,
istration numbers yielded a maximum recruitment best corrected visual acuity worse than 20/30, his¬
pool of 5940 current Barbados residents and 1660 tory or family history of major eye diseases, or the
ineligible names. The main reason for ineligibility inability to have fundus photographs taken were re¬
was nonresidence in Barbados (n=1041; 63%); other ferred for a comprehensive ophthalmologic exami¬
reasons were death (n=472; 28%) and unavailability nation, repeated tonometry, and perimetry. To as¬
for examination because of institutionalization or sess possible false-negative results, every 10th

major disability (n=147; 9%). participant was also referred for these additional
The names selected were placed in random or¬ evaluations, regardless of examination findings. The
der into a computerized recruitment system, and in¬ same ophthalmologist examined all referred partici¬
dividuals were contacted in that order to ascertain pants at the polyclinic. The examination provided
their willingness to participate in the BES. The com¬ data on the pathologic findings in the eye, the causes
puterized system was used to generate recruitment of visual acuity or field loss, and clinical diagnoses.
letters and lists of persons who needed to be called It included, for every participant, a slit-lamp exami¬
or visited at home or who required transportation, to nation of the anterior segment, an evaluation of lens
assist in the scheduling of appointments, and to opacities, and a fundus examination with an indirect
monitor participation statistics. Recruitment letters lens (Volk 78, Volk Optical, Mentor, Ohio) during
were signed by the chief medical officer of the Min¬ the slit-lamp examinations. Three-mirror gonios-
istry of Health; nonrespondents received additional copy and other evaluations were also included to es¬
mailings, telephone calls, or home recruitment vis¬ tablish the clinical diagnosis. Participants with sys¬
its. Home examinations were offered to persons who temic and/or ocular abnormalities or glycosylated
were unable to visit the study site. hemoglobin levels greater than 8% were referred
Participants attended the Sir Winston Scott elsewhere for medical evaluation and treatment.
Polyclinic, Ministry of Health, for an extensive stan¬
dardized examination that included automated re¬ PERIMETRY SCREENING PROTOCOL
fraction (Humphrey automated refractor, Humphrey
Instruments Ine, San Leandro, Calif); best corrected Evaluation of the sensitivity and specificity of vari¬
visual acuity (Ferris Bailey charts, following a modi¬ ous visual field tests during the pilot study and the
fied Early Treatment Diabetic Retinopathy Study pretesting phase of the BES led to the following
protocol)8; automated perimetry (Humphrey field screening sequence.12 All participants had a supra-
analyzer, Humphrey Instruments); two blood pres¬ threshold screening test using the full field 120 pro¬
sure measurements (Hawksley random zero sphyg- gram of the Humphrey field analyzer, three-zone
momanometer); height, weight, and circumferences strategy, in the central 30° only (4 minutes per eye);

Downloaded From: http://archopht.jamanetwork.com/ by a GLASGOW UNIVERSITY LIBRARY User on 08/01/2013

 those having one or more absolute defects (sensitivity, sidered in these definitions. The protocols to deter¬
91%; specificity, 60%) also had a threshold test with mine the glaucoma classification for each participant
the C24-2 program (12 minutes per eye). Using a com¬ are outlined below.

puter next to the perimeter, the C24-2 data were im¬ All visual fields were analyzed using computer-
mediately analyzed by the hemimeridional compari¬ based criteria to assess abnormalities. In addition, all
son method,13 using the ICEPACK program.14 If the visual field tests were reviewed and classified by one
analysis showed low sum or defects in the compari¬ ophthalmologist, who evaluated the possible artifacts
sons of field sectors 5 vs 6 and 7 vs 8 (sensitivity, 91%; and provided a clinical interpretation of the results
specificity, 64%), these individuals were referred for (normal, suspect glaucoma, definite glaucoma, cata¬
an ophthalmologic examination and a threshold C30-2 ract, other diagnosis, uncertain diagnosis, unreliable
test, which was also analyzed by the hemimeridional or missing/incomplete fields). Optic disc abnormali¬
method.13 To evaluate possible false-negatives, every ties were independently assessed by graders at the
10th person was referred to complete the full perim- fundus photography reading center and also were as¬
etry protocol. sessed by ophthalmologic examination. The ophthal¬
mologic evaluation of possible causes for visual field
OPTIC DISC EVALUATION and/or optic disc damage was sought through exami¬
nation of the participants by the BES ophthalmolo¬
Two independent graders at the reading center clas¬ gist; for those not completing this examination, the
sified the photographs, and discrepancies were re¬ ophthalmologist conducted a confirmatory review of
solved by consensus or adjudication by a third the available records.
grader. In addition to the photographic quality, rel¬ All data from the data collection center and the
evant disc features graded were horizontal and verti¬ fundus photography reading center were sent to the
cal cup-disc ratio using a template,15 narrowest re¬ coordinating center for processing, analyses, and in¬
maining neuroretinal rim, hemorrhages, and dependent glaucoma classification. Because partici¬
unsupported vessels. The ophthalmologic examina¬ pants were a cross section of the population and a
tion form included the same optic disc items as the wide spectrum of OAG stages would be included, it
reading center forms, thus allowing comparisons of was expected to have incomplete visual field or pho¬
clinical and photographic data. A quality assurance tographic data (eg, because of advanced visual loss or
system evaluated the intergrader and intragrader re- media opacities) in older participants with OAG and
producibility, as well as the reproducibility of the in those with advanced OAG. The exclusion of these
consensus photographic gradings of different grader persons would result in a biased underestimate of the
pairs. The system involved 120 sets of quality con¬ true OAG prevalence; on the other hand, the inclu¬
trol photographs that were graded first at baseline sion of persons with missing data could also lead to
and then on an ongoing basis, the masked insertion misclassification. To address this issue and define
of a quality control set after every sixth set of study minimum standards for data completeness, each of
photographs, and the assignment of these quality the BES criteria had two subcategories to specify the
control sets to different grader pairs. The reproduc¬ type of data on which the glaucoma classification was
ibility of the consensus gradings was monitored by based. For simplicity, these subcategories were desig¬
repeated evaluations of the quality control photo¬ nated with double plus signs (+ + ) to mean most
graphs by two different grading teams consisting of complete data and with a single plus sign ( + ) to
different grader pairs (eg, the consensus of graders A mean less complete data but sufficient for glaucoma
and was compared with the consensus of graders classification; they are defined in Table 1.
C and D, and the consensus of graders A and C with All participants were independently classified at
that of graders and D). Possible drift in gradings the coordinating center according to these study cri¬
over time was monitored by comparing the results of teria. Prevalence is reported both as observed per¬
various grading cycles of the quality control photo¬ centage prevalence, a minimum estimate based on
graphs. the number of cases identified in participants with
complete study visits, and as adjusted percentage
CRITERIA FOR GLAUCOMA CLASSIFICATION prevalence. For the latter, results were age-sex ad¬
justed to account both for refusals and incomplete
The BES classification of definite OAG required the study visits and for nonresponse to referrals to the
presence of both visual field and optic disc criteria in ophthalmologic examination. The adjustment as¬
at least one eye after ophthalmologic exclusion of sumed the same rate of disease in those completing
narrow angles, other types of glaucoma, and other and not completing the data collection protocol
possible causes. If these criteria were only partially within each age-sex stratum. Age-adjusted rates were
met, individuals were classified as having suspect calculated by the direct adjustment method, using a
OAG. The intraocular pressure (lOP) was not con- combined standard population.16

Downloaded From: http://archopht.jamanetwork.com/ by a GLASGOW UNIVERSITY LIBRARY User on 08/01/2013

 The Barbados Eye Study Group Table 1. Criteria for Open Angle Glaucoma (OAG)
Classification in the Barbados Eye Study (BES)*
Coordinating Center Visual field criteria
State University of New York-Stony Brook: M. Cris¬
++ At least two abnormal visual field tests by Humphrey
tina Leske, MD, MPH (Principal Investigator); Le¬ automated perimetry on two or more occasions; abnormal
slie Hyman, PhD; Roger Grimson, PhD; Edward P. tests were defined by positive results of hemimeridional
analyses of C24-2 or C30-2 threshold tests and/or the
McManmon, MPH; Ho Cheung, MS; Suh-Yuh Wu, presence of one or more absolute defects in the central 30°,
MA; Vito Squicciarini, MS; Kangjinn Peng, MA; Bar¬ as tested by the full field 120 suprathreshold test, three-zone

bara Springhorn; Kasthuri Sarma. strategy.

+ If two Humphrey visual field tests could not be completed
because of the inability to perform the tests, there was one
Data Collection Center abnormal suprathreshold Humphrey test as defined above,
and/or there was evidence of glaucomatous visual field loss
Ministry Health, Bridgetown, Barbados, West In¬
of
such as blindness or severe visual impairment.
dies; Anthea M. S. Connell, FRCS, FRCOphth; disc criteria
Coreen Barrow; Doreen Boyce; Audley Byer; Yolande
Optic
++ At least two signs of optic disc damage were present in
Babb; Anne Bradshaw;Jillia Bird, MS, OD (1989 to fundus photographs and/or the ophthalmologic examination,
1991); Valda Griffith (1988 to 1991); Hermes Nurse including either a horizontal or a vertical cup-disc ratio of
0.7 or more, narrowest remaining neuroretinal rim of 0.1
(1991 to 1992); Judith Hall (1991 to 1992); Carol disc diameters or less, notching, asymmetry in cup-disc
Selleck (1991 to 1992). ratios between eyes of more than 0.2, and disc
hemorrhages.
+ If photographs were not available, the BES ophthalmologic
FundusPhotography Reading Center examination or other clinical records documented
Thejohns Hopkins University, Baltimore, Md: An¬ glaucomatous optic nerve damage.
drew P. Schachat, MD; Judith A. Alexander; Nor- Ophthalmologic criteria
een B. Javornik, MS; Cheryl J. Hiner; Deborah A. ++ Clinical OAG diagnosis after examination by the BES

ophthalmologist. This diagnosis was based on assessment

Phillips; Reva Ward; Gregory Whitehead; Terry W. of disc and field changes, not on intraocular pressure, after
George. excluding other causes.
+ Confirmation of previous clinical OAG diagnosis and
treatment through record review (performed by the BES
ophthalmologist) for participants who did not complete the
toobtain the data needed to plan strategies to prevent or BES ophthalmologic examination.
control visual loss. Planning for the BES included the *
Double plus sign indicates most complete data; single plus sign, less
completion of a pilot study (supported by a National Eye complete data but sufficient for OAG classification.
Institute [Bethesda, Md] small grant) that tested feasi¬
bility and protocols for recruitment and data
collection.6·7 Table 2. Percentage of Participation in the
The purpose of this report is to present the BES meth¬ Barbados Eye Study by Age and Sex
ods and results regarding the prevalence of OAG. Future
Age Male, Female, Total,
reports will present data on OAG risk factors and other Group, y_No. (%)_No. (%)_No. (%)
BES aims. 40-44 315(87.3) 354(91.7) 669(89.6)
45-49 296(87.8) 368(89.1) 664(88.1)
RESULTS 50-54 224(82.7) 333(854) 557(84.3)
55-59 242(81.5) 345(87.1) 587(84.7)
PARTICIPATION 60-64 210(79.8) 351(84.0) 561(82.4)
65-69 228(844) 305(83.6) 533(83.9)
By the end of the data collection period, 4709 persons 70-74 224(79.7) 291(81.7) 515(80.8)
had BES visits and 931 refused, for a participation rate 75-79 167(76.6) 201(764) 368(76.5)
of 83.5% among the 5640 persons included in the BES 80-86 110(77.5) 145(694) 255(72.6)
Total 2016(82.5) 2693(84.3) 4709(83.5)
at the conclusion of the study in May 1992. Lack of
time or interest were the major reasons given for re¬
fusal. Participation was higher at younger than at
older ages and ranged from 89.6% at ages 40 through participants both with the total eligible sample and
44 years to 72.6% at ages older than 80 years (P<.01) with the 1990 census (Eric Straughn, director of the
(Table 2). Overall, participation was similar by sex Barbados Statistical Service, written communication,
but tended to be higher for women at younger ages. July 14, 1993). These comparisons showed similar dis¬
Because not all eligible persons in the sample were in¬ tributions; results by parish of residence are in the
cluded in the BES, possible participation biases were Figure.
evaluated by comparing the age, sex, and residence of Of the 4709 BES participants, 29 (0.6%) had

Downloaded From: http://archopht.jamanetwork.com/ by a GLASGOW UNIVERSITY LIBRARY User on 08/01/2013

 DATA COLLECTION FOR GLAUCOMA
CLASSIFICATION

Overall, 4400 persons (95%) completed at least one Hum¬

phrey visual field test; of these, 1982 (45%) had one test
and 2418 (55%) had two or more tests (Table 3). Re¬
liability indexes, as defined by Humphrey criteria,17 showed
that 2% of the participants had false-positive errors and
14% had false-negative errors of greater than 0.33; 25%
had fixation losses greater than 0.2, and 11% had fixa¬
tion losses greater than 0.3.
Fundus photography was completed by 4393 par¬
Percentage distribution of participants and Barbados 1990 census by ticipants (95%) in at least one eye, and 3929 of them (89%)
parish of residence. had gradable discs; therefore, photographic disc data were
available for 85% of the total BES population (Table 3).
Table 3. Completeness of Data for Glaucoma
Media opacities and small pupils were the main reasons
Classification in the Barbados Eye Study (N=4631) for nongradability. Baseline evaluations of reproducibil-
ity showed good agreement among the different grading
Data Item_No. (%) teams, with the highest values18 for the narrowest re¬
Humphrey perimetry 4400 (95) maining neuroretinal rim size ( =0.90) and for horizon¬
One test 1982 tal and vertical cup-disc ratios ( =0.86 and 0.88). The
Two tests 1660 results of the quality assurance scheme found no evi¬
Three or more tests 758 dence of drift. High intraclass correlations were also found
Gradable discphotographs 3929 (85) between the photographic and the clinical cup-disc ra¬
Ophthalmologic examination 2781 (93*)
Photographic or clinical disc data 4512
tio gradings (r=.87 to .89).
(97) Of the 4631 participants, 2978 (64%) were referred
'Percentage of those referred. for ophthalmologic examination, which was com¬
an

pleted by 2781 (93%) of those referred. Photographic

and/or clinical disc data were available for 97% of the par¬
Table 4. Distribution of Criteria for Open Angle Glaucoma
(OAG) Among All Persons with Definite OAG in the ticipants (Table 3).
Barbados Eye Study (n=309)
GLAUCOMA CLASSIFICATION RESULTS
Visual Field Optic Disc Examination
In the BES population, 309 persons (6.7%) of those ex¬
amined (95% confidence interval [CI], 6.3 to 7.8) met
the criteria for OAG. An additional 169 persons (3.6%)
were classified as having suspect OAG; 34 persons (0.7%)
had secondary or other types of glaucoma. Age-sex ad¬
justment for refusals and incomplete study visits re¬
sulted in an adjusted prevalence of 7.1% (95% CI, 6.5 to
*See Table 1 for description of the criteria. Double plus sign indicates 7.8) for OAG in the total BES population and of 3.7% (95%
most complete data; single plus sign, less complete data. CI, 3.2 to 4.3) for suspect OAG. When definite and sus¬
pect OAG were considered jointly, the adjusted overall
prevalence was 10.8% (95% CI, 10.0 to 11.7). Because of
incomplete study visits and 49 (1.0%) had home the high percentage of completed ophthalmologic ex¬
examinations. A previous clinical diagnosis of aminations, adjustment for nonresponse to referrals for
OAG and treatment for glaucoma was confirmed in this examination had a minimal effect on the results.
six participants who were examined at home. The
OAG analyses are based on the remaining 4631 CHARACTERISTICS OF OAG AND
participants who completed the BES protocol SUSPECT OAG CASES
at the study site. Their median age was 58 years
(mean±SD, 59±12 years), and 57% of this study The vast majority of definite OAG cases met the BES cri¬
population were female; 93% reported their race as teria designated with a double plus sign ( 4- 4- ), which re¬
black, 4% as mixed, and 3% as white or other, thus quired the most complete data, for at least two of the three
closely resembling the demographic composition of criteria (Table 4). Thus, the presence of glaucoma vi¬
the country. sual field loss was documented in at least two Hum-

Downloaded From: http://archopht.jamanetwork.com/ by a GLASGOW UNIVERSITY LIBRARY User on 08/01/2013

 Table 5. Characteristics of Persons With Open Angle Table 6. Sex and Self-reported Race of Persons
Glaucoma (OAG), With Suspect OAG, and Without (40-86 y) Classified as Having Open Angle
Glaucoma in the Barbados Eye Study Glaucoma (OAG), Suspect OAG, or Other
Glaucoma in the Barbados Eye Study (BES)
OAG Suspect OAG No Glaucoma*
(n=309) (n=169) (n=3542) Male, Female, Total,
No. (%) No. (%) No. (%)
Age, y
Mean±SD 69 i 10 62±11 57±12 All BES Participants
Median 72 63 55 Total No. 1992 2639 4631
Male sex, % 52 57 42 Open angle glaucoma 161(8.1) 148(5.6) 309(6.7)
Race, % Suspect OAG 97 (4.9) 72 (2.7) 169 (3.6)
Black 97.7 95.3 92.5 Other glaucoma 18(0.9) 16(0.6) 34(0.7)
Mixed 1.9 4.1 4.1 Black Only
White/other 0.3 0.6 3.4 Total No. 1840 2474 4314
Glaucoma treatment, % 49 11 0 Open angle glaucoma 155(8.4) 147(5.9) 302 (7.0)
IOP, mm Hgt Suspect OAG 94(5.1) 67 (2.7) 161 (3.7)
Mean ± SD 27±9 24±6 17±2 Other glaucoma 17(0.9) 13(0.5) 30 (0.7)
Median 25 24 17
Mixed Only
HCDR+.
Total No. 91 93 184
Mean ± SD 0.7±0.2 0.5±0.1 0.3±0.2
Median 0.7 0.5 0.3
Open angle glaucoma 6 (6.6) 0 6(3.3)
Suspect OAG 2 (2.2) 5 (5.4) 7 (3.8)
VCDR§ Other glaucoma 0 1(1.1) 1 (0.5)
Mean±SD =0.1 0.6±0.1 0.3±0.2
Median 0.6 0.3 White/Other Only
Total No. 61 72 133
"Excludes 34 persons with other types of glaucoma and 577 with Open angle glaucoma 0 1(1.4) 1(0.8)
intraocular pressure (IOP) >21 mm Hg.
"[Highest average IOP in either eye. Suspect OAG 1(1.6) 0 1(0.8)
^Highest horizontal cup-disc ratio (HCDR) in either eye. Other glaucoma 1(1.6) 2(2.8) 3(2.3)
^Highest vertical cup-disc ratio (VCDR) in either eye.

phrey visual field tests for most of the cases (groups 1,3, sons who were classified as having definite OAG, sus¬
5, and 6; Table 4). The rest were unable to complete two pect OAG, or other glaucoma are presented in Table 6.
tests, mainly because of severe visual impairment. Disc The OAG prevalence by self-reported race was 7.0% (302/
damage was documented by photographic and/or clini¬ 4314) in black, 3.3% (6/184) in mixed-race, and 0.8%
cal assessment in most cases (over four fifths) as well (1/133) in white participants (Table 6). All six partici¬
(groups 1, 2, and 5; Table 4). All but 12 cases were ex¬ pants in the mixed-race group were male.
amined by the BES ophthalmologist (groups 1 through
4; Table 4). AGE-SEX SPECIFIC PREVALENCE
Persons with OAG and suspect OAG were signifi¬
cantly older, and there was a higher percentage of male The results that follow are based on the 4498 black and
and black participants than in the nonglaucoma popu¬ mixed-race participants only. The observed overall preva¬
lation (Table 5). The median age of the participants with lence of OAG in this group was 6.8% (95% CI, 6.1 to 7.6)
OAG was 72 years for both men and women. Overall, the and varied significantly by age and sex. It increased steeply
median age of the participants with suspect OAG was 63 with age, reaching 14.8% at ages 70 through 79 years and
years; it was 62 years for men and 65 years for women. 23.2% at older ages (Table 7). In every age group, preva¬
Of the participants with OAG, 49% had received previ¬ lence was higher in men than in women, with an age-
ous diagnoses of OAG and were receiving treatment. There adjusted prevalence ratio of 1.4.
were no differences in the age and sex distribution of par¬ An additional 168 persons (3.7%) (95% CI, 3.2 to
ticipants with newly diagnosed and previously diag¬ 4.3) were classified as having suspect OAG (Table 7). As
nosed OAG. About 11% of those with suspect OAG re¬ in OAG, the prevalence of suspect OAG was higher in
ported treatment for glaucoma but did not meet the study men than in women; however, it remained relatively stable
criteria for OAG. The average IOP was elevated in the with age (6% to 7% in men and 2% to 4% in women). If
OAG and the suspect OAG groups. High cup-disc ratios these suspect OAG cases are added to the definite OAG
were present in participants with OAG and suspect OAG cases, the combined prevalence is 10.6% (95% CI, 9.7 to
as well (Table 5). 11.5), with 13.3% of the men and 8.5% of the women
The sex distribution and self-reported race of per- affected (Table 7).

Downloaded From: http://archopht.jamanetwork.com/ by a GLASGOW UNIVERSITY LIBRARY User on 08/01/2013

 Table 7. Age-Sex Specific Distribution of Open Angle Glaucoma (OAG)
and Suspect OAG in 4498 Barbados Eye Study Participants*

Male Female Total

Group No. (%) 95% CI No. (%) 95% CI No. (%) 95% CI
OAG
40-49 11(1.9) 0.9-3.3 7(1.0) 0.4-2.0 18(1.4) 0.8-2.2
50-59 21 (4.6) 2.9-7.0 24(3.7) 2.4-5.4 45 (4.1) 3.0-5.4
60-69 40 (9.4) 6.8-12.6 31 (4.9) 3.4-6.9 71 (6.7) 5.3-8.4
70-79 64(17.5) 13.8-21.8 58(12.7) 9.8-16.1 122(14.8) 12.5-17.4
80-86 25 (24.8) 16.7-34.3 27(22.0) 15.0-30.3 52 (23.2) 17.9-29.3
Total 161 (8.3) 7.1-9.7 147 (5.7) 4.9-6.7 308 (6.8) 6.1-7.6

Suspect OAG
40-49 13(2.2) 1.2-3.8 14 (2.0) 1.1-3.3 27 (2.2) 1.4-3.1
50-59 26 (5.8) 3.8-8.3 14(2.1) 1.2-3.6 40 (3.7) 2.7-5.0
60-69 26(6.1) 4.0-8.8 22 (3.5) 2.2-5.2 48 (4.7) 3.5-6.2
70-79 24 (6.6) 4.3-9.6 19(4.2) 2.9-6.4 43 (5.5) 4.0-7.2
80-86 7(6.9) 2.8-13.8 3 (2.4) 0.5-7.0 10(4.0) 1.8-7.5
Total 96 (5.0) 4.1-6.0 72 (2.8) 2.2-3.5 168 (3.7) 3.2-4.3
OAG and Suspect OAG
Total 257(13.3) 11.8-14.9 219 (8.5) 7.5-9.7 476 (10.6) 9.7-11.5

"Excludes 133 persons self-reported as white/other. CI indicates confidence interval.

COMMENT identified in any population-based study,20"29 thus en¬

hancing its ability to detect risk factors. To avoid the in¬
To our knowledge, the BES is the largest and most com¬ clusion of questionable cases of OAG, the BES preva¬
prehensive investigation of OAG and other major eye dis¬ lence results were based on a conservative and strict
eases in a black population. The study had a high par¬ definition of OAG that required the presence of both vi¬
ticipation rate (Table 2), and the BES population had a sual field and optic disc criteria. As such, it may be in¬
demographic composition similar to the census popula¬ terpreted as a minimum prevalence based on cases with
tion. Because the BES was based on a large, representa¬ definite and established disease as determined by stan¬
tive sample of the adult resident population, its results dardized testing at the study site. Because persons who
may be generalized to the country as a whole. The study only partially met the BES criteria were classified as hav¬
was designed to provide much needed data for public ing suspect OAG, this group includes individuals who
health planning, as well as to investigate the cause of OAG were considered to have OAG by other definitions, eg,
and other visually impairing diseases in people of Afri¬ persons with glaucomatous disc damage who did not meet
can descent. Data were collected according to standard the BES criteria for visual field loss. Therefore, compari¬
protocols, objective methods were used to define and grade sons of BES findings with those of other studies must con¬

the diseases of interest, and quality control of all data was sider that both visual field and optic disc criteria were
carefully monitored. The fundus photographs were read required for the definition of OAG in the BES. Field loss
by masked graders, and the reproducibility of clinical and and optic disc damage were documented by automated
photographic protocols was evaluated throughout the perimetry, fundus photography, and a standardized oph¬
study. Perimetric reliability was consistent with other re¬ thalmologic examination for most participants with OAG
ports,17 and the sensitivity of the screening protocol was (Table 4). The main reasons for having less complete data
high.12 Despite the difficulties of performing automated were the presence of advanced visual loss and the inabil¬

perimetry, fundus photography, and other tests in a gen¬ ity to complete perimetry or fundus photography.
eral population, a high degree of data completeness was The number of BES participants who reported be¬
achieved (Table 3). ing white or of mixed black and white ancestry was too
As reviewed in detail elsewhere,L·19 most OAG popu¬ small to draw conclusions from, but the results were con¬
lation studies have been limited to white adults, in whom sistent with an increased prevalence of OAG in the black
prevalence has been 1% or less. Therefore, the number participants, as evidenced by the comparisons with the
of prevalent cases identified by these studies has been in¬ nonglaucoma group (Table 5). A gradient in prevalence
sufficient for extensive evaluations. To our knowledge, by racial group is also suggested (Table 6).
the BES included the largest number of OAG cases ever Half of the OAG cases identified in the BES were

Downloaded From: http://archopht.jamanetwork.com/ by a GLASGOW UNIVERSITY LIBRARY User on 08/01/2013

 newly detected by the study (Table 5), a finding that is OAG (4%) among 2395 urban black residents and 32
consistent with the results of other prevalence studies in (1%) among 2913 whites; it had 75% to 79% partici¬
the United States and Europe.20·30·31 A comparison of newly pation. Even if the probable cases in this survey are
diagnosed and previously diagnosed cases found no age added the definite cases, there seems to be a lower
to
or gender differences between these groups. prevalence of OAG in the Baltimore study than in the
Among those with OAG, half had IOPs of 25 mm BES. Interpretation of these results must consider
Hg or more in at least one eye (Table 5). High median some differences between the studies. The Baltimore

cup-disc ratios of 0.8 were also found in this group (Table study used the full field 120 program for screening
5), a result that is consistent with the BES criteria for OAG. and had different perimetry criteria for ophthalmo¬
Of the participants with suspect OAG, who may repre¬ logic referral. The diagnosis of OAG was based on
sent participants with early OAG, 11% were receiving treat¬ Goldmann perimetry, rather than on automated Hum¬
ment for glaucoma, and they had higher average IOPs phrey perimetry. Despite these differences, however, it
and cup-disc ratios than did the nonglaucoma group seems reasonable to conclude that a higher prevalence

(Table 5). does occur in Afro-Caribbean than in Afro-American

populations. Age-adjusted comparisons show that the
BES prevalence of OAG in black participants is 7.1

Analyses
that were limited to black and times higher than that observed in the Baltimore Eye
mixed-race BES participants show the Survey white participants and 1.5 times higher than
high prevalence of OAG with increas¬ that observed in the Baltimore Eye Survey black par¬
ing age (Table 7). In the 40- through 49- ticipants. Discrepancies between the prevalences from
year age group, the observed prevalence the BES and from studies in the United States and Eu¬
was similar to that observed in much older white popu¬ rope are even larger when BES participants with sus¬
lations.20 24 Large increases are seen in successive age pect OAG are included in these comparisons. These
groups; over half the participants were over the age of 70 differences raise the question of possible genetic fac¬
years, in whom the prevalence reaches about 15%. An tors and require further study.
even higher prevalence of 23% was seen in the oldest age The distribution patterns of OAG in the BES
group. Other studies in the Caribbean region with simi¬ showed a preponderance of OAG in men (Tables 5
larly high participation rates have also found a higher through 7). Some studies1·21·27 have reported higher
prevalence of OAG than that in white populations. In Ja¬ rates of glaucoma blindness and a higher prevalence of
maica, West Indies, a study25 with 85% participation re¬ OAG in men than in women (eg, the Framingham Eye
ported a prevalence of 1.4% (9/678), which was three times Study had a twofold higher prevalence in men than in
higher than that in a comparable study in Wales.20 A study women).21 This finding has not been consistently re¬
in St Lucia, West Indies, reported an 8.8% prevalence of ported by other studies in white populations, but re¬
OAG at ages 30 years and older among 1679 black resi¬ sults are based on few cases of OAG. Data from the
dents screened.26 previous smaller studies in blacks found similar sex-
Although the results of the BES and the St Lucia study specific prevalences of OAG. However, the sex-specific
are consistent with a high prevalence of OAG in both is¬ data from St Lucia are difficult to interpret because of
lands, the studies differ in methods and definitions. The the underrepresentation of the male population, ie,
St Lucia protocol did not include fundus photographs. only 32% of their participants were male vs 44% in the
Visual field screening with the full field 120 test was pro¬ population.26 As the authors suggest, the unwilling¬
vided to every third person and to those meeting criteria ness of men to participate led both to an underenu-
based on IOP and clinical assessment of the cup-disc ra¬ meration of men (36%) in the household survey used
tio; 364 persons (70% of those referred) had a C30-2 test. as a sampling frame and to their low participation in
The 8.8% prevalence was based on 147 cases with posi¬ examinations. The Baltimore study did not report sex-
tive hemimeridional analyses after ophthalmologic re¬ specific participation for blacks, but the age- and race-
view of the 410 eyes with reliable C30-2 tests. Preva¬ adjusted prevalence of OAG was similar in both sexes
lence was 16% when based on a secondary definition. The (2.70% for men and 2.35% for women).23
apparently higher age-specific prevalence in the St Lucia Possible biases that would result in differential re¬
study than in the BES could be explained by the differ¬ cruitment of men with OAG into the BES were ex¬
ences in OAG protocols and diagnostic criteria, eg, the plored. The sex-specific participation in the BES
requirement for optic disc plus visual field criteria in the tended to be somewhat higher in women than in men
BES. For example, age-specific prevalences are similar in and would not explain the differences in prevalence.
both studies when the BES definite and suspect cases of There was no evidence of a higher participation of
OAG are combined. There also may be some differences men than women because of their previous diagnosis
between populations. of OAG, because the percentage of cases with a known
The Baltimore Eye Survey23 identified 100 cases of history of glaucoma was similar in both sexes. Simi-

Downloaded From: http://archopht.jamanetwork.com/ by a GLASGOW UNIVERSITY LIBRARY User on 08/01/2013

 betic Retinopathy Study: Manual of Operations. Baltimore, Md: ETDRS Coor-
larly, family history of glaucoma was reported more dinating Center, Dept of Epidemiology and Preventive Medicine, University of
frequently in women than in men. The BES results are Maryland School of Medicine; 1980.
thus consistent with an increased male prevalence of 11. The GHb Test: True Glycosylated Hemoglobins by Affinity Chromotography.
OAG in this black population, a finding that merits Akron, Ohio: ISOLAB; 1989.
further exploration. 12. Oden N, Leske MC, Connell A, Troutman T, the Barbados Eye Study Group.
The BES is the largest glaucoma study conducted in Using automated perimetry to screen for open-angle glaucoma in a black popu-
lation. Invest Ophthalmol Vis Sci. 1992;33(abstract issue):757. Abstract.
a black population and identified more persons with OAG 13. Sommer A, Enger C, Witt K. Screening for glaucomatous visual field depth with
than any previous population study. The BES results con¬ automated threshold perimetry. Am J Ophthalmol. 1987;103:681-684.
firm the public health impact of OAG in black people, 14. Canner JK, Sommer A, Katz J, Enger C. ICEPACK: A user friendly software
package for processing humphrey field analyzer diskettes. Invest Ophthalmol
especially at older ages and in men. Extrapolation of BES Vis Sci. 1988;29(suppl):356.
results to the country's population indicates that one in 15. Klein B, Moss SE, Magli YL, et al. Optic disc cupping as clinically estimated
11 persons older than 50 years has OAG; this estimate from photographs. Ophthalmology. 1987;94:1481-1483.
increases to one in nine at ages over 60 years and to one 16. Kleinbaum DG, Kupper LL, Morgenstern H. Epidemiologic research: principles
in six in those over 70 years. and quantitative methods. New York, NY: Van Nostrand Reinhold; 1982.
17. Katz J, Sommer A. Screening for glaucomatous visual field loss: the effect of
patient reliability. Ophthalmology. 1990;97:1032-1037.
Accepted for publication March 1, 1994. 18. Fleiss JL. Statistical Methods for Rates and Proportions. 2nd ed. New York,
Supported by grants EY07625 and EY07617 from the 19.
NY: Wiley & Sons; 1981.
National Leske MC, Rosenthal J. Epidemiologic aspects of open-angle glaucoma. Am J
Eye Institute, Bethesda, Md. Epidemiol. 1979;109:259-272.
Reprint requests to the Department of Preventive Medi¬ 20. Hollows FC, Graham PA. Intraocular pressure, glaucoma and glaucoma sus-
cine, Schoolof Medicine, University of New York-Stony pects in a defined population. Br J Ophthalmol. 1966;50:570-586.
Brook, HSC L2 154, Stony Brook, NY 11794-8275 (Dr 21. Kahn HA, Leibowitz HM, Ganley JP, et al. The Framingham Eye Study, I: outline
and major prevalence findings. Am J Epidemiol. 1977;106:17-32.
Leske).
22. Bengtsson B. The prevalence of glaucoma. Br J Ophthalmol. 1981;65:46-49.
23. Tielsch JM, Sommer A, Katz J, Royall RM, Quigley HA, Javitt J. Racial varia-
REFERENCES tions in the prevalence of primary open-angle glaucoma: the Baltimore Eye
Survey. JAMA. 1991;266:369-374.
1. Leske MC. The epidemiology of open-angle glaucoma: a review. Am J Epide- 24. Klein B, Klein R, Sponsel WE, et al. Prevalence of glaucoma: the Beaver Dam
miol. 1983;118:166-191. Study. Ophthalmology. 1992;99:1499-1504.
2. US Dept of Health and Human Services. Vision Research\p=m-\ANational Plan 25. Wallace J, Lovell HG. Glaucoma and intraocular pressure in Jamaica. Am J
1994-1998: A Report of the Glaucoma Panel. Bethesda, Md: National Advisory Ophthalmol. 1969;67:93-100.
Eye Council; 1993. Publication NIH 93-3186. 26. Mason RP, Omofolasade K, Wilson MR, et al. National prevalence and risk
3. Hiller R, Kahn HA. Blindness from glaucoma. Am J Ophthalmol. 1975;80:62-69. factors of glaucoma in St Lucia, West Indies. Ophthalmology. 1989;96:1363\x=req-\
4. Coulehan JL, Helzlsouer KJ, Rogers KD, Brown SI. Racial differences in intra- 1368.
ocular tension and glaucoma surgery. Am J Epidemiol. 1980;111:759-768. 27. Viggosson G, Bjornsson G, Ingvason JG. The prevalence of open-angle glau-
5. Sommer A, Tielsch JM, Katz J, et al. Racial differences in the cause-specific coma in Iceland. Acta Ophthalmol (Copenh). 1986;64:138-141.

prevalence of blindness in East Baltimore. N Engl J Med. 1991;125:1442-1447. 28. Shiose Y, Kitazawa Y, Tsukahara S, et al. Epidemiology of glaucoma in Japan:
6. Leske MC, Connell AMS. Design of a pilot study of glaucoma in Barbados. a nationwide glaucoma survey. Jpn J Ophthalmol. 1991;35:133-155.

J Natl Med Assoc. 1988;80:727-730. 29. Salmon JF, Mermoud A, Ivey A, Swanevelder SA, Hoffman J. The prevalence
7. Leske MC, Connell AMS, Kehoe R. A pilot project of glaucoma in Barbados. Br of primary angle closure glaucoma and open angle glaucoma in Mamre, West-
J Ophthalmol. 1989;73:365-369. ern Cape, South Africa. Arch Ophthalmol. 1993;111:1263-1268.
8. Ferris FL, Kassoff A, Bresnick GH, Bailey I. New visual acuity charts for clinical 30. Leske MC, Podgor M, Ederer F. An evaluation of glaucoma screening methods.
research. Am J Ophthalmol. 1982;94:91-96. Invest Ophthalmol Vis Sci. 1982;22(suppl):128.
9. Chylack LT Jr, Leske MC, McCarthy D, Khu P, Kashiwagi T, Sperduto R. Lens 31. Sommer A, Tielsch JM, Katz J, et al. Relationship between intraocular pressure
Opacities Classification System II (LOCS II). Arch Ophthalmol. 1989;107:991-997. and primary open angle glaucoma among white and black Americans: the Bal-
10. The Early Treatment Diabetic Retinopathy Study Group. Early Treatment Dia- timore Eye Survey. Arch Ophthalmol. 1991;109:1090-1095.

Downloaded From: http://archopht.jamanetwork.com/ by a GLASGOW UNIVERSITY LIBRARY User on 08/01/2013