Risk Factors for Incident Open-angle

Glaucoma
The Barbados Eye Studies
M. Cristina Leske, MD, MPH,1,2 Suh-Yuh Wu, MA,1,2 Anselm Hennis, FRCP(UK), PhD,1,3,4
Robert Honkanen, MD,2 Barbara Nemesure, PhD,1,2 BESs Study Group

Purpose: To evaluate risk factors for definite open-angle glaucoma (OAG), based on African-descent
participants of the Barbados Eye Studies.
Design: Cohort study with 81% to 85% participation over 9 years’ follow-up.
Participants: We evaluated 3222 persons at risk, 40 to 84 years old, who did not have definite OAG at
baseline.
Methods: Participants had standardized study visits at baseline and after 4 and 9 years, with structured
interviews, blood pressure (BP), and other measurements. The ophthalmic protocol included automated perim-
etry, applanation tonometry, fundus photography, and comprehensive ophthalmologic examinations for those
referred. Central corneal thickness (CCT) was measured in a subset at the 9-year examination. Incidence was
estimated by the product-limit approach; relative risk ratios (RRs) with 95% confidence intervals (CIs) were based
on Cox regression models with discrete time.
Main Outcome Measure: Nine-year incidence of definite OAG.
Results: Over 9 years, 125 persons developed definite OAG (incidence, 4.4%; 95% CI, 3.7–5.2). Baseline
factors influencing risk were age (RR, 1.04; 95% CI, 1.02–1.05 per year); family history of glaucoma (RR, 2.4; 95%
CI, 1.3– 4.6); higher intraocular pressure (IOP) (RR, 1.12; 95% CI, 1.08 –1.16 per mmHg); lower systolic BP (RR,
0.91; 95% CI, 0.84 –1.00 per 10 mmHg); and lower ocular systolic, diastolic, and mean perfusion pressures (RR,
0.66; 95% CI, 0.54 – 0.80 per 10 mmHg higher mean perfusion pressure) (RR, 2.6; 95% CI, 1.4 – 4.6 for low mean
perfusion pressure [⬍40 mmHg]). Thinner CCT was also associated with OAG incidence (odds ratio, 1.41; 95%
CI, 1.01–1.96 per 40 ␮m lower).
Conclusions: This is the first report of risk factors for long-term OAG incidence; it is also based on a sizable
number of new cases. Incidence was high in this African-descent population, where the established factors of
older age, higher IOP, and family history contributed to risk. Additional predictors were vascular factors, including
lower systolic BP, and particularly lower ocular perfusion pressures, which more than doubled risk. Thinner CCT
was also a factor. These findings indicate a multifactorial etiology of OAG and suggest that similar risk factors
apply across populations. Results are relevant for understanding OAG causation and identifying groups at high
risk. Ophthalmology 2008;115:85–93 © 2008 by the American Academy of Ophthalmology.

Despite extensive research over many years, the causal
events leading to open-angle glaucoma (OAG) are not well
understood. Epidemiologic studies have made considerable
contributions, not only by determining OAG frequency in
various populations, but also by identifying risk factors for
the disease. Available evidence has firmly established older
age, family history, intraocular pressure (IOP), and African
ancestry as risk factors, but data on other potential factors
are inconsistent. Current information was derived mainly
from cross-sectional data obtained by prevalence-based or
case-control studies, because very few prospective cohort
studies of OAG incidence have been conducted to date.1–5
Yet incidence studies are most valuable to determine risk
factors. Their longitudinal design provides a direct measure

4
Originally received: October 19, 2006. Chronic Disease Research Centre, Tropical Medicine Research Institute,
Final revision: March 9, 2007. University of the West Indies, Bridgetown, Barbados.
Accepted: March 9, 2007. 5
Wilmer Institute, The Johns Hopkins University, Baltimore, Maryland.
Available online: July 16, 2007. Manuscript no. 2006-1195.
1
Supported by the National Eye Institute, Bethesda, Maryland (grant nos.
Department of Preventive Medicine, Stony Brook University School of EY07625, EY07617).
Medicine, Stony Brook, New York.
2
Correspondence to M. Cristina Leske, MD, MPH, Department of Preven-
Department of Ophthalmology, Stony Brook University School of Med- tive Medicine, Stony Brook University School of Medicine, Health Sci-
icine, Stony Brook, New York. ences Center, L3 086, Stony Brook, NY 11794-8036. E-mail: cleske@
3
Ministry of Health, Bridgetown, Barbados. notes.cc.sunysb.edu.

© 2008 by the American Academy of Ophthalmology ISSN 0161-6420/08/$–see front matter 85

Published by Elsevier Inc. doi:10.1016/j.ophtha.2007.03.017

Downloaded for Ryan Juliansyah (ryan.naturalis@gmail.com) at Tarumanagara University from ClinicalKey.com by Elsevier on September 24, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.
 Ophthalmology Volume 115, Number 1, January 2008
of the true risk of developing new disease over a time
period; it also allows determining which preexisting factors,
present at baseline, are related to the subsequent risk of new
disease.
One issue limiting OAG incidence studies has been the
low frequency of new cases per year in most populations,
with small sample sizes and insufficient statistical power to
identify some risk factors. Prospective studies in black
populations can overcome this limitation, given the high
frequency of OAG in persons of African descent. We have
reported on the 4- and 9-year incidence of OAG in the
Barbados Eye Studies (BESs), which were designed to
provide long-term data on all major eye diseases in a pop-
ulation with the same ancestry as African Americans.4,5
During the 9-year follow-up period, 125 persons newly
developed definite OAG, which is the largest number of
population-based incident cases reported to date. The cur-
rent article aims to evaluate risk factors for OAG, based on
these 9-year incidence data.
Materials and Methods
Background
The major aims of the BESs were to determine prevalence,
incidence, progression, and risk factors for the main eye dis-
eases in Barbados, West Indies, a country where most persons
(⬎90%) are of African descent. The background and methods
of this National Eye Institute-funded study have been reported
previously.4 –7 In summary, the BESs include a cohort of 4709
persons, identified through a simple random sample of the
country’s population, 40 to 84 years of age, with 84% partici-
pation. By self-report, the ancestry of the 4631 persons exam-
ined at the study site was 93% (n ⫽ 4314) African/black, 4%
(n ⫽ 184) mixed (black and white), and 3% (n ⫽ 133) Euro-
pean/white or other.6 After an initial baseline prevalence phase,
the Barbados Eye Study (1987–1992), the surviving cohort
members were reexamined 4 and 9 years later in the Barbados
Incidence Study of Eye Diseases I (1992–1997; 85% participa-
tion),4 and II (1997–2002; 81% participation).7 As compared
with participants, nonparticipants in the 9-year examination
were older and more likely to have hypertension and higher IOP
at baseline; the deceased group contained more men and had
more frequently reported a history of diabetes at baseline.7,8
The study was reviewed and approved by the institutional
review boards of collaborating institutions and informed con-
sent was obtained from all study participants.
Data for Open-angle Glaucoma Classification
The study visits followed the same standardized protocols, with
extensive data collection. Protocols relevant to this report in-
cluded a structured interview on demographic, ocular, medical
and other risk factor data; various measurements, including 2
blood pressure (BP) readings with a random zero sphygmoma-
nometer, various ophthalmic measurements, including Hum-
phrey perimetry (suprathreshold C64 program of the Humphrey
Visual Field Analyzer, 3-zone strategy for all participants;
full-threshold C24-2 and C30-2 tests for those meeting criteria
for abnormal tests)4,5; applanation tonometry; central corneal
thickness (CCT) measured at the 9-year visit on a sample of
1142 consecutive participants (KMI ultrasonic RK5000
pachymeter; KMI Surgical, Paoli, PA9); color stereo fundus
86
Downloaded for Ryan Juliansyah (ryan.naturalis@gmail.com) at Tarumanagara University from ClinicalKey.com by Elsevier on September 24, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.
photographs; and a comprehensive ophthalmologic examination
with dilatation, repeat tonometry, and perimetry (full-threshold
tests) for those referred and a 10% sample. All visual field tests
were evaluated by the study’s ophthalmologists to provide a
clinical interpretation. Fundus photographs of the disc and
macula were independently assessed at the study’s reading
center (at the Wilmer Institute, The Johns Hopkins University,
Baltimore, MD), with good reproducibility throughout the
study. Data completeness for the glaucoma detection protocols
was very high at all visits, with 89% successfully completing
the visual field tests and 97% having photographic and/or
clinical disc gradings at the 9-year visit.4 – 6
As described elsewhere,5 an algorithm was used for the
glaucoma classification, with independent review by a glau-
coma specialist (RH) at the Coordinating Center. The classifi-
cation of definite OAG required the presence of glaucoma
visual field defects and optic disc damage in at least 1 eye, after
ophthalmologic exclusion of other possible causes, without
considering IOP. The criteria for definite glaucoma visual field
defect specified ⱖ2 abnormal visual field tests by Humphrey
perimetry (based on hemimeridional analyses with glaucoma
hemifield tests) with ophthalmologic interpretation as definite
or suspect glaucomatous field loss, or ⬍2 abnormal tests with
ophthalmologic interpretation as definite glaucomatous field
loss. The criteria for glaucomatous optic neuropathy specified
ⱖ2 signs of optic disc damage (horizontal and vertical cup-to-
disc ratio ⱖ0.7, narrowest remaining neuroretinal rim ⱕ0.1 disc
diameters, notching, asymmetry in cup-to-disc ratios between
eyes ⬎0.2, or disc hemorrhages) in the fundus photographs
graded at the study’s reading center and/or the ophthalmologic
examination, or ⬍2 signs with an ophthalmologic assessment or
clinical record documenting definite glaucomatous optic nerve
damage.4 – 6,10
Risk Factors
The factors evaluated were demographic (age, gender, education,
occupation), medical history (e.g., diabetes, hypertension, cardio-
vascular disease), anthropometric and BP-related measurements
(body mass index, systolic BP [SBP] and diastolic BP [DBP],
hypertension [SBP ⱖ 140 mmHg, DBP ⱖ 90 mmHg, or antihy-
pertensive treatment], ocular systolic perfusion pressures [SPP],
diastolic perfusion pressure [DPP], and mean perfusion pressure
[MPP; SPP ⫽ SBP ⫺ IOP, DPP ⫽ DBP ⫺ IOP, MPP ⫽ 2/3 mean
arterial pressure ⫺ IOP], arterial pressure [DBP ⫹ 1/3 (SBP ⫺
DBP)], and pulse pressure [SBP ⫺ DBP]), family history (glau-
coma family history in parents or siblings), and ocular factors (IOP
[average of 3 applanation measurements] and CCT [average of 5
ultrasound measurements], based on the affected eye for unilateral
incident cases and the worse eye for bilateral incident cases).
Statistical Methods
The person-based 9-year cumulative incidence of definite OAG
was defined as the development of OAG in either eye during
follow-up, based on persons without definite OAG in either eye
at baseline. After excluding persons with non-African ancestry
(because of small sample size), or with bilateral secondary and
other types of glaucoma, 4008 participants of African ancestry
remained. Of these, 3222 (80.4%) completed at least 1 follow-up
examination and were considered the population at risk for these
analyses. The cumulative incidence rates were estimated by the
product-limit approach11; relative risk ratios (RR) with 95% con-
fidence intervals (CI) were based on Cox regression models with
discrete time.12 Because CCT data were not available on all
participants, the CCT analyses followed a case-control within a
 Leske et al 䡠 Risk Factors for Incident Open-angle Glaucoma: The Barbados Eye Studies

Table 1. Baseline Characteristics of Persons at Risk (n ⫽ 3222) also analyzed as categorical variables using quartile and quintile
cutoff values based on the distribution of the study population. The
Age (yrs), mean ⫾ SD (median) 56.9⫾11.3 (56.0) latter categories were further grouped into low, medium, and high,
Male 41.0% with values above the upper quintile as high, those below the lower
Family history of glaucoma 6.7% quintile as low, and others as medium.
Diabetes history 16.7%
Hypertension 52.4%
BP (mmHg)
Systolic, mean ⫾ SD (median) 135.3⫾22.6 (132.0) Results
Diastolic, mean ⫾ SD (median) 81.2⫾12.0 (80.0)
PP (mmHg)
Systolic, mean ⫾ SD (median) 117.3⫾22.2 (114.7) As shown in Table 1, the 3222 persons at risk of developing
Diastolic, mean ⫾ SD (median) 63.3⫾12.2 (62.7) definite OAG at baseline had a median age of 56.9 years, 41%
Mean, mean ⫾ SD (median) 48.2⫾9.5 (47.7) were male, and 7% had a family history of glaucoma. Diabetes and
IOP (mmHg), mean ⫾ SD (median) 18.0⫾4.1 (17.7) hypertension were common, at 17% and 52%, respectively. The
IOP-lowering treatment 1.6% median IOP was 17.7 mmHg and the median CCT (among the
Corneal thickness (␮m), mean ⫾ SD (median)* 538.4⫾37.5 (537.0) subset with pachymetry) was 537 ␮m. Table 1 also provides
descriptive data on the BP and ocular perfusion pressure values in
BP ⫽ blood pressure; IOP ⫽ intraocular pressure; PP ⫽ perfusion pressure; this population.
SD ⫽ standard deviation. Over the 9-year follow-up, 125 persons developed definite
*n ⫽ 1023, measured at the 9-yr visit. glaucoma, for an incidence of 4.4% (95% CI, 3.7–5.2), as previ-
ously reported; of these, 67 cases had developed by the 4-year
follow-up.5 Table 2 presents the RR for demographic, familial, and
cohort approach, which compared CCT measurements of incident other risk factors. Increasing age was strongly related to OAG risk,
cases with the remaining participants. These analyses were based with a 4% increase in the RR per year of age (RR, 1.04 per year).
on logistic regression models for CCT and presented as odds ratios As compared with persons 40 to 49 years of age, the RR was
(OR) with 95% CI. Age- and gender- adjusted RRs and ORs were ⬎2-fold for those 60 to 69 years and 2.6-fold for those ⱖ70 years
first evaluated and significant variables (Pⱕ0.05) were subse- and older (RR, 2.3 and 2.6, respectively). Although men had a
quently retained in multivariable models adjusting for age, gender, higher incidence than women, the RR was not statistically signif-
IOP, and IOP- and BP-lowering treatments. Continuous data were icant; educational status was unrelated to risk.

Table 2. Demographic, Familial and Medical History Factors, Intraocular Pressure (IOP), and Risk of
Definite Open-angle Glaucoma

Factor n 9-Year Incidence Adjusted RR (95% CI)* P Value

Age (yrs) 1.04 (1.02–1.05) 0.0001
40–49 1060 2.2 (1.5–3.4) 1.0 —
50–59 878 3.6 (2.5–5.1) 1.4 (0.8–2.5) 0.22
60–69 742 6.6 (4.9–8.9) 2.3 (1.3–3.9) 0.003
ⱖ70 542 7.9 (5.6–11.1) 2.6 (1.5–4.6) 0.001
Gender
Female 1901 4.1 (3.2–5.1) 1.0 —
Male 1321 4.9 (3.8–6.4) 1.2 (0.8–1.7) 0.32
Education (yrs)
Low (⬍9) 557 6.4 (4.5–9.1) 1.0 —
Medium (9–11) 1771 4.3 (3.4–5.5) 0.9 (0.6–1.4) 0.62
High (ⱖ12) 793 3.1 (2.1–4.7) 0.8 (0.5–1.6) 0.65
Glaucoma family history
No 2996 4.1 (3.4–4.9) 1.0 —
Yes 216 8.9 (5.7–13.7) 2.4 (1.3–4.6) 0.002
Diabetes history
No 2675 4.2 (3.5–5.1) 1.0 —
Yes 536 5.8 (3.9–8.6) 1.2 (0.7–1.8) 0.49
Hypertension
No 1529 3.8 (2.9–5.0) 1.0 —
Yes 1685 5.1 (4.0–6.4) 0.8 (0.5–1.2) 0.26
Blood pressure–lowering treatment
No 2294 4.2 (3.4–5.1) 1.0 —
Yes 918 5.1 (3.8–7.0) 1.0 (0.7–1.5) 0.96
Cataract history
No 2901 4.2 (3.5–5.1) 1.0 —
Yes 307 6.3 (3.8–10.2) 0.9 (0.5–1.6) 0.66
IOP per mmHg 1.12 (1.08–1.16) 0.0001

RR ⫽ relative risk.
*Based on Cox regression models adjusting for age, gender, IOP, and IOP and blood pressure treatment; education,
cataract history, and diabetes history were adjusted for age and gender.

87

Downloaded for Ryan Juliansyah (ryan.naturalis@gmail.com) at Tarumanagara University from ClinicalKey.com by Elsevier on September 24, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.
 Ophthalmology Volume 115, Number 1, January 2008

Table 3. Blood Pressure (BP)–Related Factors and Risk of Definite Open-angle Glaucoma

Systolic BP Diastolic BP Pulse Pressure Arterial Pressure

†
Per 10-mmHg increase 0.91 (0.84–1.00)* 0.92 (0.79–1.08) 0.90 (0.80–1.01) 0.89 (0.77–1.02)
High‡ 1.0 1.0 1.0 1.0
Medium‡ 0.8 (0.5–1.3) 1.0 (0.6–1.7) 1.2 (0.7–1.9) 0.8 (0.5–1.3)
Low‡ 1.3 (0.7–2.4) 1.3 (0.7–2.3) 1.2 (0.6–2.3) 1.2 (0.7–2.2)
Q4‡ 1.0 1.0 1.0 1.0
Q3‡ 1.1 (0.7–1.8) 1.0 (0.6–1.7) 0.9 (0.5–1.6) 0.8 (0.5–1.3)
Q2‡ 0.9 (0.5–1.6) 0.9 (0.5–1.5) 1.5 (0.9–2.6) 0.9 (0.6–1.6)
Q1‡ 1.4 (0.8–2.4) 1.4 (0.8–2.3) 0.9 (0.5–1.8) 1.4 (0.8–2.4)

SPP DPP MPP

Per 10-mmHg increase 0.87 (0.79–0.95)* 0.79 (0.68–0.92)* 0.66 (0.54–0.80)*
High§ 1.0 1.0 1.0
Medium§ 1.1 (0.7–1.8) 1.3 (0.7–2.2) 1.2 (0.7–2.1)
Low§ 2.00 (1.1–3.5)* 2.1 (1.2–3.9)* 2.6 (1.4–4.6)*
Q4§ 1.0 1.0 1.0
Q3§ 1.0 (0.6–1.7) 1.1 (0.6–2.1) 1.1 (0.6–2.0)
Q2§ 1.2 (0.7–2.1) 1.5 (0.8–2.7) 1.4 (0.8–2.5)
Q1§ 1.9 (1.1–3.3)* 2.2 (1.3–3.8)* 2.2 (1.3–3.8)*

DPP ⫽ diastolic perfusion pressure; MPP ⫽ mean perfusion pressure; SPP ⫽ systolic perfusion pressure.
Relative risks (95% confidence intervals) based on Cox regression models adjusting for age, gender, intraocular pressure (IOP), and IOP-lowering and
BP-lowering treatment.
*P ⫽ 0.05.
†
P ⫽ 0.08.
‡
Systolic BP (mmHg): low, ⱕ116; high, ⬎153; Q1, ⱕ119; Q2, 119 –132; Q3, 132–148; Q4, ⬎148. Diastolic BP (mmHg): low, ⱕ71; high, ⬎90; Q1, ⱕ73;
Q2, 73– 80; Q3, 80 – 88; Q4, ⬎88. Pulse pressure (mmHg): low, ⱕ40; high, ⬎67; Q1, ⱕ41; Q2, 41–51; Q3, 51– 64; Q4, ⬎64. Arterial pressure (mmHg):
low, ⱕ87; high, ⬎110; Q1, ⱕ89; Q2, 89 –98; Q3, 98 –107; Q4, ⬎107.
§
SPP (mmHg): low, ⱕ98; high, ⬎134; Q1, ⱕ101; Q2, 101–115; Q3, 115–130; Q4, ⬎130. DPP (mmHg): low, ⱕ53; high, ⬎73; Q1, ⱕ55; Q2, 55– 63; Q3,
63–70; Q4, ⬎70. MPP (mmHg): low, ⱕ40; high, ⬎55; Q1, ⱕ42; Q2, 42– 48; Q3, 48 –54; Q4, ⬎54.

Glaucoma family history increased risk almost 2.5 times (RR, risk with thinner CCT was also observed in all the categorical anal-
2.4). The age- and gender-adjusted RR for hypertension was 0.8, yses. The OR was significantly associated with a 3-fold likelihood of
suggesting a reduced risk, but was not statistically significant. No OAG (OR, 3.1) in participants with CCTⱕ505 ␮m.
associations were found with antihypertensive treatment or cata- Figure 1 provides a summary of the significant risk factors for
ract history. The IOP was an additional major factor, with a 12% incidence of definite OAG, based on the final multivariable
increase in RR per mmHg higher of IOP (RR, 1.12 per mmHg). As models presented in Tables 2 through 4. The figure shows
reported previously,13 there was a considerable increase in inci- decreased RR for SBP (RR, 0.91 per 10-mmHg higher) and
dence with higher levels of baseline IOP. increased RR for older age (RR, 1.04 per year), glaucoma
Table 3 provides data on BP-related factors and OAG. As BP family history (RR, 2.4), IOP (RR, 1.12 per mmHg higher), low
increased, the risk of OAG tended to decrease. A 10-mmHg
ocular MPP ⬍ 40 mmHg; RR, 2.6), and thinner CCT (OR, 1.41
increase in SBP was associated with a 9% decrease in risk (RR,
per 40-␮m thinner CCT).
0.91), with very similar results observed for DBP, pulse pressure,
and arterial pressure, all with RR⬍1. A parallel but nonsignificant
trend was observed when pressure values were grouped into cat-
egories, with most RRs increasing in magnitude as the SBP, DBP, Table 4. Association of Corneal Thickness (CCT) and
and pulse and arterial pressures decreased. Incident Open-angle Glaucoma
Consistent with these results, lower ocular perfusion pressures
were associated with risk. All the RRs for lower SPP, DPP, and CCT n OR (95% CI)*
MPP were statistically significant, whether derived from analyses
that used continuous or categorical variables. For example, there Per 40 ␮m thinner 1023 1.41 (1.01–1.96)†
was a 34% average decrease in the RR per 10 mmHg increase in High (⬎562 ␮m) 204 1.0
Medium (501–562 ␮m) 613 2.4 (0.9–6.7)
MPP (RR, 0.66), with consistent results for SPP and DPP (RR,
Low (ⱕ501 ␮m) 206 2.8 (0.9–9.1)
0.87 and 0.79, respectively). A similar pattern was observed for
Q4 (⬎555 ␮m) 255 1.0
analyses of perfusion pressure categories, as the RR increased with Q3 (530–555 ␮m) 255 2.6 (0.9–7.4)
low pressure values; for example, the RR for participants with low Q2 (505–530 ␮m) 256 3.0 (1.1–8.7)†
DPP (⬍53 mmHg) was 2.2 and it was 2.6 for participants with low Q1 (ⱕ505 ␮m) 257 3.1 (1.1–8.8)†
MPP (⬍40 mmHg).
Table 4 displays the results of analyses for associations with
CCT, which are presented as OR based on logistic regression, CI ⫽ confidence interval; OR ⫽ odds ratio.
using a similar format as Table 3. As CCT decreased, OAG risk *Based on logistic regression models adjusting for age, gender, intraocular
pressure (IOP), and IOP-lowering and blood pressure-lowering treatment.
significantly increased, with an approximately 40% higher likelihood †
P⬍0.05.
of OAG per 40-␮m-thinner CCT (OR, 1.41). A trend toward a higher

88

Downloaded for Ryan Juliansyah (ryan.naturalis@gmail.com) at Tarumanagara University from ClinicalKey.com by Elsevier on September 24, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.
 Leske et al 䡠 Risk Factors for Incident Open-angle Glaucoma: The Barbados Eye Studies

Figure 1. Risk factors for definite open-angle glaucoma (OAG; n ⫽ 3222). hx ⫽ history; PP ⫽ perfusion pressure; RR ⫽ risk ratio; SBP ⫽ systolic blood
pressure. *Based on Cox regression models, adjusting for age, gender, intraocular pressure (IOP), and IOP- and blood pressure–lowering treatment; central
corneal thickness (CCT) is presented as an odds ratio, based on logistic regression model in a subsample (n ⫽ 1023).

Discussion Demographic Factors

After a follow-up of 9 years, we found a high OAG
incidence of 4.4% in this African-descent population,
averaging about 0.5% per year.5 In comparison, data
from European-derived populations, although based on a
small number of incident cases, suggest rates of 0.1% to
0.3% per year.1–3 What is the explanation for the higher
risk observed? Aside from a true difference in the mag-
nitude of the risk, possible explanations include differ-
ences in OAG classification criteria, data collection
methods, or other design issues. Another possibility
could be the inclusion of persons with suspect OAG in
our population at risk, which would have led to an
increase in the incidence rate. After evaluating these
possibilities and their potential impact, as discussed in
detail elsewhere,5 we believe that our African-descent
population has a truly higher risk. This conclusion is in
line with a metaanalysis of prevalence studies, which
confirmed differences in prevalence by ancestry.14
The population-based design, longitudinal follow-up,
and large sample size of our study facilitated the assessment
of risk factors for incident, definite OAG. Because OAG
incidence is equivalent to OAG risk, a major advantage is
that results have easier interpretation than results of cross-
sectional studies, which do not measure risk and are sus-
ceptible to biases. The results of this cohort study confirmed
the well-known factors of older age, higher IOP, and family
history (Table 2), which appear constant across populations.
Additional findings were decreased RRs for BP variables,
which were also detected in our 4-year incidence analyses,
along with associations with lower perfusion pressures (Ta-
ble 3).15 A new risk factor for this population was thinner
CCT (Table 4), with the caveat of being measured at 9
years, not at baseline. The sections that follow discuss these
results and their implications.
Increasing age is a major risk factor for OAG, as the
disease seldom occurs at ages ⬍40 years. Our results
suggest that risk more than doubles after age 60, as
compared with the 40- to 49-year-old age group. Age also
increased risk in the other available prospective cohort
studies conducted in Australia and Europe.2,3,16 Although
incidence was higher among men, the RR was marginally
significant, which could be due to insufficient sample
size. In fact, there was a significant positive association
with male gender (RR, 1.31; 95% CI, 1.02–1.67) when
we combined definite and suspect/probable glaucoma
(i.e., persons meeting some, but not all, of our strict OAG
criteria).5 A trend toward an increased risk in males was
also found in the incidence studies from Melbourne,
Australia,2 and Rotterdam, The Netherlands,3 but not
from Dalby, Sweden.1 All these studies were based on a
small number of definite incident cases (n ⫽ 10, 29, and
26, respectively) in European-derived populations. For
prevalence studies, a metaanalysis14 found that men were
more likely to have OAG than women, with an OR of
1.37 (95% CI, 1.22–1.53), which is similar to the RR of
our incidence results. Taken together, the prevalence and
incidence data suggest that males have increased risk of
definite OAG, but additional incidence studies are needed
to confirm these results. No other demographic factors
were associated with OAG in our analyses.
Family History
As previously reported, a history of OAG in the family was
an important risk factor.4,6 Although associations with fa-
milial history are consistent across OAG studies, no clear
genetic patterns have emerged. Reporting and other biases
are always an issue when assessing family history, as shown
by the higher frequency of such history in persons who are
aware, compared with those unaware, of their OAG diag-

89

Downloaded for Ryan Juliansyah (ryan.naturalis@gmail.com) at Tarumanagara University from ClinicalKey.com by Elsevier on September 24, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.
 Ophthalmology Volume 115, Number 1, January 2008
Figure 2. Hypertension (defined as ⱖ140/90 mmHg or antihypertensive
treatment), age, and risk of definite open-angle glaucoma. RR ⫽ risk ratio.
**Based on Cox regression models.
nosis.17–19 Yet the unaware OAG cases in these studies still
reported familial history more frequently than persons with-
out OAG. Furthermore, examinations of relatives of OAG
probands have confirmed they have an increased frequency
of the disease.10,20 In the Barbados Family Study of Open-
angle Glaucoma, definite or suspect/probable OAG was
found in a high percentage (almost 40%) of the relatives of
OAG probands,10 with a likely codominant mode of trans-
mission21 and with linkage to intervals on chromosomes 2q
and 10p.22 It is likely that ancestral genetic factors contrib-
ute to the high OAG risk in this and other African-derived
populations, with the magnitude of the risk varying with the
distribution of these factors. For example, the higher risk in
African Caribbeans than African Americans could be due to
different degrees of admixture. Continued research is
needed to unravel the genetic component of OAG in various
populations, which may depend on gene– environment
interactions.
Medical History
Hypertension and diabetes are often mentioned as OAG risk
factors, but the supporting evidence is not strong, with
inconsistent reports. For hypertension, epidemiologic stud-
ies have reported positive, negative, or no significant rela-
tionships.15,23–28 Most of the available evidence originates
from cross-sectional studies, which do not allow firm con-
clusions on risk factors. Whereas data from cohort studies
are better to assess the true relationships of hypertension
and OAG, if any, the results either find no relationships16 or
indicate a reduced risk.15 Hypertension was of special in-
terest as a potential risk factor in our study, given its high
frequency in this and other black populations. Consistent
with the decreased RR for SBP and DBP, the RR for
hypertension was also decreased (RR, 0.8), suggesting that
hypertensive patients tended to have lower OAG risks.
These results are comparable to our previous 4-year data,15
but the RR was not statistically significant in the 9-year
data. The role of hypertension in OAG could be age related,
with hypertension being protective at younger ages and
increasing risk at older ages.17 This possibility was explored
and results are presented in Figure 2, which shows RRs for
hypertension in various age groups, as well as overall. All
90
Downloaded for Ryan Juliansyah (ryan.naturalis@gmail.com) at Tarumanagara University from ClinicalKey.com by Elsevier on September 24, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.
the RRs were decreased and not statistically significant,
indicating no age effect, with the lowest RR, of 0.5, in the
oldest age group. Similar results were observed in a Mexi-
can American population, where hypertension was not as-
sociated with OAG and no age effect was apparent.23 As
found in other studies,26 the use of systemic antihyperten-
sive medications was unrelated to OAG.
The relationships with diabetes are also inconsistent;
whereas a metaanalysis of cross-sectional studies suggested
a positive relationship, recent incidence-based data indi-
cated no link between diabetes and OAG.29 –33 Our results
are consistent with the latter report. We also found no
relationship with cataract history.
Intraocular Pressure
The link between IOP and OAG is well established, and was
again confirmed in our study. Persons with IOP ⬎21 or ⬎25
mmHg at baseline had a markedly elevated risk of OAG
(adjusted RR, 5.2 [95% CI, 3.5–7.6] and 5.9 [95% CI,
3.6 –9.4], respectively)13; however, relatively few persons
in the population had such pressures and most of the 125
new cases developed among persons with lower IOP. In
fact, 54% of the incident OAG cases had IOPⱕ21 mmHg at
baseline. A more extensive discussion of the IOP results can
be found in another publication.13
Blood Pressure and Ocular Perfusion Pressure
Evaluations of SBP, DBP, pulse pressure, and arterial pres-
sure consistently indicated a trend toward negative relation-
ships with OAG risk. The RRs per mmHg were all de-
creased for these four variables (0.89 – 0.92 per 10 mmHg
higher). When data were grouped as high, medium, and low,
the lowest categories of SBP and DBP had RR⬎1, in the
direction of an increased risk; results were not as uniform
when SBP and DBP were categorized by quartiles, perhaps
indicating a minor role of BP, by itself, on OAG risk. A
similar pattern was found for arterial pressure and for pulse
pressure, which has been found to be an independent car-
diovascular risk factor.34 All these results are consistent
with the reduced RR found for hypertension. As a whole,
our data suggest a weak association between low BP and
OAG risk. This inference is in agreement with a recent
review, which concludes that evidence is stronger for a link
between OAG and hypotension than for hypertension.35 The
hypothesized mechanism is that low BP compromises ocu-
lar perfusion pressure at the optic disc and thus leads to
glaucomatous damage. In a recent study, lower BP and
lower perfusion pressure were related to increased optic disc
cupping and thinner neuroretinal rim in persons without
glaucoma,36 thereby strengthening the possibility of a vas-
cular link to glaucoma.37
All lower ocular perfusion pressures in our study were
positively related to OAG risk. The lower the perfusion
pressure, the higher the risk, with RR at least doubling in the
lowest perfusion pressure categories. Similar results were
found in our 4-year data, as well as in various other epide-
miologic studies, where low perfusion pressures are a con-
sistent finding, as opposed to the variable associations with
 Leske et al 䡠 Risk Factors for Incident Open-angle Glaucoma: The Barbados Eye Studies
BP.10,15,23–27 As summarized in various reviews,38 – 40 lower
perfusion pressures at the optic disc would compromise
ocular blood flow and thus lead to glaucoma damage.
Although the associations with lower perfusion pressures
are biologically plausible and are in concert with the hy-
pothesized mechanism, their interpretation is not clear-cut.
In addition to reflecting low BP, such associations could be
influenced by high IOP alone or by treatment to reduce the
IOP or the BP. These variables were controlled for in our
analyses, based on IOP and medication data obtained at
study visits, but residual confounding is always possible.
However, we also found decreased RR for BP variables, and
no associations with treatment, supporting the conclusion
that lower perfusion pressures are the key risk factor. Ocular
perfusion pressures, which also depend on IOP, may be
more pertinent to OAG risk than BP alone. In our popula-
tion, which has high average IOP and high average BP
(Table 1), it may be necessary to maintain those BP levels
to protect the disc. Therefore, even moderately low BPs
could be deleterious by causing low ocular perfusion pres-
sures and potentially increasing OAG risk. The BP–IOP
interrelationships may be different across populations and
result in different associations. Focusing on BP alone as a
risk factor has limitations, as brachial pressure does
not reflect the pressure status at the optic disc, which is most
relevant. Perfusion pressure thus seems a more pertinent
factor.
Central Corneal Thickness
Thinner CCT was related to incident OAG (Table 4). To our
knowledge, this is the first report of such finding in a
population-based cohort study. These results agree with
those of the Ocular Hypertension Study,41 which found a
strong association between CCT and OAG development in
a clinical trial of ocular hypertension. As in the Ocular
Hypertension Study, the CCT measurements in our study
were not obtained at baseline, which is a limitation to
consider when interpreting the results. Furthermore, these
measurements were available only for a subset of our par-
ticipants, thus lowering the power of the risk factor analy-
ses. The CCT varies according to ancestry and has been
shown to be thinner in African-derived populations.9,42 In
previous cross-sectional analyses, we also found low CCT
values in our participants; there was a marginally significant
relationship (P ⫽ 0.07) between CCT and OAG,9 now
confirmed by the current incidence data. The thinner central
corneas of black populations may be an ancestrally related
factor that additionally contributes to the increased OAG
risk in these populations.
As recently reviewed,43 the mechanisms for a CCT–
OAG association have yet to be determined. A thin central
cornea may lead to an underestimation of true IOP, causing
errors in measured IOP values. Possibly, CCT could be an
indicator of the biomechanical and structural characteristics
of ocular tissues, which may have a strong influence on
OAG risk. Eyes with thinner CCT thus could have increased
elasticity or a particular property that renders the optic
nerve tissue more susceptible to damage than other eyes.
Downloaded for Ryan Juliansyah (ryan.naturalis@gmail.com) at Tarumanagara University from ClinicalKey.com by Elsevier on September 24, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.
Thin central corneas could increase OAG risk through this
mechanism.
Strengths, Limitations, and Generalizability
This longitudinal study has the major strengths of determin-
ing risk factors for incidence, as opposed to prevalence, and
of being population based. In addition, results are based on
incident definite OAG, using a rigorous definition, and did
not include any probable or suspect cases. When we re-
peated the analyses based on incident cases that included
persons with suspect/probable OAG, similar results were
found. As noted, male gender became a significant factor in
these analyses, but we opted for presenting the more con-
servative findings. Study data were collected according to
carefully designed protocols, which were standardized and
aimed at high data quality (e.g., use of random zero sphyg-
momanometer) and completeness. Although we were able
to achieve good follow-up over an extended time period,
attrition owing to mortality or other causes is always an
issue, as in all cohort studies. Losses to follow-up could lead
to biases, for example, because of the higher likelihood of
hypertension among nonparticipants. One limitation in in-
terpreting the CCT results, as mentioned, is that pachymetry
was performed in a sample at the 9-year examination. As
such, our results are based on a cross-sectional analysis;
assumptions must be made regarding the stability of CCT
over time and its value as a predictor. Regarding general-
izability of results, one issue to consider is that the study is
based on a black population, which has the same ancestral
origin as African Americans.44 This population has free
access to health care with a high life expectancy of
about 76 years45 and, despite differences in environmen-
tal and admixture factors, has similar morbidity and
mortality patterns as African Americans, particularly re-
garding hypertension.46,47
In conclusion, after 9 years of follow-up, a high inci-
dence of OAG was found in this population of African
origin. Although ancestral genetic factors are likely to un-
derlie the high risk, glaucoma development appears to be a
multifactorial process. In fact, our analyses confirmed the
important contribution of several factors. The results sug-
gest that etiologic factors are generally similar across pop-
ulations, because the same factors increased risk in studies
from other geographic areas and very different settings. As
such, the magnitude of OAG risk in a given population, for
example, European, Asian, or African derived, may depend
on the frequency of the pertinent factors in that population.
As in other studies, age, family history of glaucoma, and
IOP were strongly related to OAG risk. Our data also
suggest a role for vascular factors. We conclude that hyper-
tension or high BP, which were common in our black
participants, do not explain their high glaucoma risk. In-
stead, results suggest a trend for associations with lower BP.
Consistent with this finding, lower ocular perfusion pres-
sures were related to OAG development, and may be more
relevant risk factors than BP per se. These results highlight
the importance of maintaining an adequate ocular perfusion
pressure in this population, which has high overall levels of
91
 Ophthalmology Volume 115, Number 1, January 2008
IOP. Thinner CCT was a new risk factor that may also play
a part in the high risk of African-descent participants.
References
1. Bengtsson BO. Incidence of manifest glaucoma. Br J Oph-
thalmol 1989;73:483–7.
2. Mukesh BN, McCarty CA, Rait JL, Taylor HR. Five-year
incidence of open-angle glaucoma: the Visual Impairment
Project. Ophthalmology 2002;109:1047–51.
3. de Voogd S, Ikram MK, Wolfs RC, et al. Incidence of open-
angle glaucoma in a general elderly population. The Rotter-
dam Study. Ophthalmology 2005;112:1487–93.
4. Leske MC, Connell AM, Wu SY, et al. Incidence of open-
angle glaucoma: the Barbados Eye Studies. Arch Ophthalmol
2001;119:89 –95.
5. Leske MC, Wu SY, Honkanen R, et al. Nine-year incidence of
open-angle glaucoma in the Barbados Eye Studies. Ophthal-
mology 2006;113:29 –35.
6. Leske MC, Connell AM, Schachat AP, Hyman L. The Bar-
bados Eye Study: prevalence of open angle glaucoma. Arch
Ophthalmol 1994;112:821–9.
7. Leske MC, Wu SY, Nemesure B, et al. Incidence and pro-
gression of lens opacities in the Barbados Eye Studies. Oph-
thalmology 2000;107:1267–73.
8. Wu SY, Nemesure B, Hennis A, et al. Nine-year changes in
intraocular pressure: the Barbados Eye Studies. Arch Ophthal-
mol 2006;124:1631– 6.
9. Nemesure B, Wu SY, Hennis A, et al. Corneal thickness and
intraocular pressure in the Barbados eye studies. Arch Oph-
thalmol 2003;121:240 – 4.
10. Leske MC, Nemesure B, He Q, Barbados Family Study
Group. Patterns of open-angle glaucoma in the Barbados
Family Study. Ophthalmology 2001;108:1015–22.
11. Kaplan EL, Meier P. Nonparametric estimation from incom-
plete observations. J Am Stat Assoc 1958;53:457– 81.
12. Cox C. Multinomial regression models based on continuation
ratios. Stat Med 1988;7:435– 41.
13. Nemesure B, Honkanen R, Hennis A, et al. Incident open-angle
glaucoma and intraocular pressure. Ophthalmology. In press.
14. Rudnicka AR, Mt-Isa S, Owen CG, et al. Variations in pri-
mary open-angle glaucoma prevalence by age, gender, and
race: a Bayesian meta-analysis. Invest Ophthalmol Vis Sci
2006;47:4254 – 61.
15. Leske MC, Wu SY, Nemesure B, et al. Incident open-angle
glaucoma and blood pressure. Arch Ophthalmol 2002;120:
954 –9.
16. Le A, Mukesh BN, McCarty CA, Taylor HR. Risk factors
associated with the incidence of open-angle glaucoma: the
Visual Impairment Project. Invest Ophthalmol Vis Sci 2003;
44:3783–9.
17. Tielsch JM, Katz J, Sommer A, et al. Family history and risk
of primary open angle glaucoma: the Baltimore Eye Survey.
Arch Ophthalmol 1994;112:69 –73.
18. Nemesure B, Leske MC, He Q, et al. Analyses of reported
family history of glaucoma: a preliminary investigation. Oph-
thalmic Epidemiol 1996;3:135– 41.
19. Mitchell P, Rochtchina E, Lee AJ, Wang JJ. Bias in self-
reported family history and relationship to glaucoma: the
Blue Mountains Eye Study. Ophthalmic Epidemiol 2002;9:
333– 45.
20. Wolfs RC, Klaver CC, Ramrattan RS, et al. Genetic risk of
primary open-angle glaucoma: population-based familial ag-
gregation study. Arch Ophthalmol 1998;116:1640 –5.
92
Downloaded for Ryan Juliansyah (ryan.naturalis@gmail.com) at Tarumanagara University from ClinicalKey.com by Elsevier on September 24, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.
21. Nemesure B, He Q, Mendell N, et al. Inheritance of open-
angle glaucoma in the Barbados Family Study. Am J Med
Genet 2001;103:36 – 43.
22. Nemesure B, Jiao X, He Q, et al. A genome-wide scan for
primary open-angle glaucoma (POAG): the Barbados Family
Study of Open-angle Glaucoma. Hum Genet 2003;112:600 –9.
23. Quigley HA, West SK, Rodriguez J, et al. The prevalence of
glaucoma in a population-based study of Hispanic subjects:
Proyecto VER. Arch Ophthalmol 2001;119:1819 –26.
24. Tielsch JM, Katz J, Sommer A, et al. Hypertension, perfusion
pressure, and primary open-angle glaucoma: a population-
based assessment. Arch Ophthalmol 1995;113:216 –21.
25. Leske MC, Connell AM, Wu SY, et al. Risk factors for
open-angle glaucoma: the Barbados Eye Study. Arch Ophthal-
mol 1995;113:918 –24.
26. Leske MC, Warheit-Roberts L, Wu SY. Open-angle glaucoma
and ocular hypertension: the Long Island Glaucoma Case-
Control Study. Ophthalmic Epidemiol 1996;3:85–96.
27. Bonomi L, Marchini G, Marraffa M, et al. Vascular risk
factors for primary open angle glaucoma: the Egna-Neumarkt
Study. Ophthalmology 2000;107:1287–93.
28. Mitchell P, Lee AJ, Rochtchina E, Wang JJ. Open-angle
glaucoma and systemic hypertension: the Blue Mountains Eye
Study. J Glaucoma 2004;13:319 –26.
29. de Voogd S, Ikram MK, Wolfs RC, et al. Is diabetes mellitus
a risk factor for open-angle glaucoma? The Rotterdam Study.
Ophthalmology 2006;113:1827–31.
30. Pasquale LR, Kang JH, Manson JE, et al. Prospective study of
type 2 diabetes mellitus and risk of primary open-angle glau-
coma in women. Ophthalmology 2006;113:1081– 6.
31. Nakamura M, Kanamori A, Negi A. Diabetes mellitus as a risk
factor for glaucomatous optic neuropathy. Ophthalmologica
2005;219:1–10.
32. Bonovas S, Peponis V, Filioussi K. Diabetes mellitus as a risk
factor for primary open-angle glaucoma: a meta-analysis. Dia-
bet Med 2004;21:609 –14.
33. Ellis JD, Evans JM, Ruta DA, et al. Glaucoma incidence in an
unselected cohort of diabetic patients: is diabetes mellitus a risk
factor for glaucoma? Br J Ophthalmol 2000;84:1218 –24.
34. Safar ME. Systolic blood pressure, pulse pressure and arterial
stiffness as cardiovascular risk factors. Curr Opin Nephrol
Hypertens 2001;10:257– 61.
35. Pache M, Flammer J. A sick eye in a sick body? Systemic
findings in patients with primary open-angle glaucoma. Surv
Ophthalmol 2006;51:179 –212.
36. Topouzis F, Coleman AL, Harris A, et al. Association of blood
pressure status with the optic disk structure in non-glaucoma
subjects: the Thessaloniki Eye Study. Am J Ophthalmol 2006;
142:60 –7.
37. Jonas JB. Association of blood pressure status with the optic
disk structure. Am J Ophthalmol 2006;142:144 –5.
38. Flammer J, Orgul S, Costa VP, et al. The impact of ocular
blood flow in glaucoma. Prog Retin Eye Res 2002;21:
359 –93.
39. Harris A, Rechtman E, Siesky B, et al. The role of optic nerve
blood flow in the pathogenesis of glaucoma. Ophthalmol Clin
North Am 2005;18:345–53, v.
40. Grieshaber MC, Flammer J. Blood flow in glaucoma. Curr
Opin Ophthalmol 2005;16:79 – 83.
41. Gordon MO, Beiser JA, Brandt JD, et al. The Ocular Hyper-
tension Treatment Study: baseline factors that predict the
onset of primary open-angle glaucoma. Arch Ophthalmol
2002;120:714 –20, discussion 829 –30.
42. Aghaian E, Choe JE, Lin S, Stamper RL. Central corneal
thickness of Caucasians, Chinese, Hispanics, Filipinos, Afri-
 Leske et al 䡠 Risk Factors for Incident Open-angle Glaucoma: The Barbados Eye Studies

can Americans, and Japanese in a glaucoma clinic.
Ophthalmology 2004;111:2211–9.
43. Brown KE, Congdon NG. Corneal structure and biomechanics:
impact on the diagnosis and management of glaucoma. Curr
Opin Ophthalmol 2006;17:338 – 43.
44. Goodridge R. The African background of the Barbados pop-
ulation. In: Cobley AG, Thompson A, eds. The African-
Caribbean Connection: Historical and Cultural Perspectives.
Cave Hill, Barbados: University of the West Indies, Depart-
ment of History; 1990:28 – 48.
45. Pan American Health Organization. Health Analysis and Infor-
mation Systems Area. Regional Core Health Data Initiative;
Technical Health Information System. Washington; 2005.
46. Holder Y, Lewis MJ. Epidemiological overview of morbidity and
mortality. In: Health Conditions in the Caribbean. Geneva:
WHO; 1997:22– 61. PAHO Scientific Publication 561.
47. Burt VL, Whelton P, Roccella EJ, et al. Prevalence of hyper-
tension in the US adult population: results from the Third
National Health and Nutrition Examination Survey, 1988 –
1991. Hypertension 1995;25:305–13.

93

Downloaded for Ryan Juliansyah (ryan.naturalis@gmail.com) at Tarumanagara University from ClinicalKey.com by Elsevier on September 24, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.