Wu S, Nemesure B, Hennis A, Schachat AP, Hyman L, Leske MC, Barbados Eye Studies Group. Open-angle Glaucoma and MortalityThe Barbados Eye Studies. Arch Ophthalmol. 2008;126(3):365-370. doi:10.1001/archophthalmol.2007.77
To evaluate the relationship between open-angle glaucoma (OAG) and mortality in a black population at 9-years' follow-up.
Population-based cohort study of 4092 black participants (aged 40-84 years at baseline) in the Barbados Eye Studies. Open-angle glaucoma was defined by visual field defects and optic disc damage, based on standardized examinations and photograph gradings. Ocular hypertension was defined by an intraocular pressure greater than 21 mm Hg or treatment, without OAG damage. Mortality was ascertained from death certificates. Cox proportional hazards regression analyses determined associations with mortality.
After 9 years, 764 (19%) participants were deceased. Mortality was unrelated to overall OAG at baseline (n = 300) after adjustment for confounders. However, cardiovascular mortality tended to increase in persons with previously diagnosed/treated OAG (n = 141; relative risk [RR], 1.38, P = .07) and was significantly higher with treatment involving timolol maleate (RR, 1.91, P = .04). Cardiovascular deaths also tended to increase in persons with ocular hypertension at baseline (n = 498; RR, 1.28, P = .06).
In this black population, cardiovascular mortality tended to increase in persons with previously diagnosed/treated OAG and ocular hypertension. The excess mortality associated with timolol maleate treatment of OAG, also found in a white population, warrants further investigation.
Glaucoma is one of the leading causes of visual impairment worldwide.1 The most common type, primary open-angle glaucoma (OAG), is especially prevalent in populations of African origin, in which it is the foremost cause of blindness.2- 5 In addition to its impact on vision and quality of life, this age-related ocular disease may have other consequences, such as an increased risk of premature mortality. Several recent studies, mainly in white populations, have attempted to determine if persons with glaucoma or high intraocular pressure (IOP) have excess mortality.6- 13 While some of these reports suggest that high IOPS are associated with decreased survival,6,7,13 others have found no association,8- 12 leading to uncertainty regarding the mortality risk of those affected with the disease.14 Information on the relationship between glaucoma and risk of death has particular clinical and public health importance in black populations, which have high rates of chronic disease mortality, higher IOPs, and much higher prevalence/incidence of OAG than white populations.2,3,15- 17 To the best of our knowledge, no long-term data exist on mortality and glaucoma/ocular hypertension in persons of African descent. The 4-year follow-up of the predominantly black participants of the Barbados Eye Studies indicated similar mortality rates for persons with and without OAG, especially among those aged 60 years or older.18 Our study extends these observations by evaluating possible relationships between OAG at baseline and subsequent mortality in black participants in this cohort after 9 years of follow-up.
The Barbados Eye Studies, funded by the National Eye Institute, are population-based investigations of the prevalence, incidence, and risk factors for major eye diseases in the predominantly black population of Barbados, which has a similar ancestry to African American individuals. The study sites included a coordinating center (Stony Brook University, Stony Brook, New York); a data collection center (Ministry of Health, Bridgetown, Barbados); and a fundus photography reading center (The Johns Hopkins University, Baltimore, Maryland). The baseline prevalence study (Barbados Eye Study, 1987-1992; 84% participation) included a simple random sample of Barbadian-born citizens aged 40 to 84 years. Of the 4631 persons examined at the study site, 4314 (93%) self-reported their race as black, 184 (4%) as mixed (black and white), and 133 (3%) as white/other.3 To determine incidence and progression, the surviving members of the cohort were reexamined 4 and 9 years after their baseline visit in the Barbados Incidence Study of Eye Diseases I15 (1992-1997; 85% participation) and Barbados Incidence Study of Eye Diseases II19 (1997-2003; 81% participation), respectively.
The standardized protocols in all study visits, as described previously,3,15,19 included anthropometric and blood pressure measurements; an interview; and various ocular measurements, such as best-corrected visual acuity, Humphrey automated perimetry, applanation tonometry, lens gradings, and fundus stereo photography. A systematic 10% sample and participants with specific findings (eg, positive ocular disease or diabetes history, abnormal perimetry, IOP > 21 mm Hg, best-corrected visual acuity < 20/30) were referred for comprehensive ophthalmologic examinations.
Definite OAG was classified by specific study criteria that required both visual field defects and optic disc damage in at least 1 eye, after excluding other possible causes; diagnoses were confirmed by the study ophthalmologists. A participant's IOP was not considered in this definition. Persons with definite OAG and prior diagnosis and treatment were categorized as having previously diagnosed/treated OAG, otherwise they were considered to have newly diagnosed/untreated OAG. Persons without OAG but with an IOP greater than 21 mm Hg (based on the highest mean value of 3 measurements for each eye) were considered to have ocular hypertension. For the purposes of this article, individuals without OAG who received IOP-lowering treatment were considered to have an IOP greater than 21 mm Hg and were thus classified as having ocular hypertension. Persons classified as having suspect/probable glaucoma (having some but not all of the strict optic disc and visual field criteria for OAG) as well as those with secondary or other glaucoma types were excluded.
Mortality data were verified from death certificates held at the Ministry of Health. Certificates included dates and specific causes of death, which were extracted and coded according to the International Classification of Diseases, Ninth Revision (ICD-9).20 Deaths that occurred during the 9-year follow-up were considered events in this article. The survival intervals were calculated as the days between the baseline visit and the Barbados Incidence Study of Eye Diseases II visit (or the last contact date for nonparticipants not reported as deceased). Analyses evaluated the associations of OAG at baseline to all-cause mortality and the most frequent cause, cardiovascular mortality. Cardiovascular disease classification was based on ICD-9 codes 390 through 448. Age-, sex-, and multivariate-adjusted relative risks (RRs) of hazards and associated 95% confidence intervals (CIs) were based on Cox proportional hazards regression analyses.21 Potential factors included simultaneously in the multivariate adjustment were age, sex, hypertension (systolic blood pressure ≥ 140 mm Hg or diastolic blood pressure ≥ 90 mm Hg and/or a history of antihypertensive treatment), and self-reported history of diabetes, cardiac disease (including angina/infarction from the Rose questionnaire22 or self-reported history of myocardial infarction), and stroke, as well as IOP and IOP-lowering treatment at baseline.
The study population consisted of 4092 of the 4314 black participants who completed baseline visits at the study site; 33 participants with unknown survival status at the 9-year follow-up, 159 persons with suspect/probable OAG, and 30 persons with secondary or other types of glaucoma were excluded. Participants' mean (SD) age at baseline was 58.6 (12.1) years and 58% were female (Table 1). Systemic diseases, such as diabetes and hypertension, were highly prevalent at baseline (18% and 55%, respectively). More than 7% of the participants were classified as having definite OAG and 4.2% of the participants received IOP-lowering treatment at baseline. The mean (SD) IOP was 18.8 (5.1) mm Hg, with 16% of participants having IOPs higher than 21 mm Hg.
Of the 4092 total participants, 764 (19%) had died by the end of the 9-year follow-up. Table 2 presents deceased participants in the cohort according to the presence of OAG and ocular hypertension at baseline as well as the use of IOP-lowering treatment. While about half of the participants with OAG were previously diagnosed and treated (141 of 300), few participants with ocular hypertension were receiving treatment (32 of 498). The observed mortality frequencies appear higher in those with OAG (37%), ocular hypertension, or treatment than those without, a finding influenced by age and other factors, as shown by further analyses.
Table 3 presents RRs for death by baseline OAG status for all-cause mortality and for cardiovascular mortality, which accounted for 53% (404) of deaths. The unadjusted RRs for OAG were significant, with approximately 2 to 3 times excess risk of all-cause and cardiovascular deaths. However, these associations were no longer significant after adjusting for age alone or age and sex, or after multivariate adjustment. The multivariate model simultaneously adjusted for a number of factors, of which age, sex, hypertension, self -reported history of diabetes, cardiac disease, and stroke were significantly associated with mortality (Table 3). Further analyses found no associations between OAG and risk of death when stratifying by age nor any significant interaction between OAG and age. Similarly, there was no significant evidence of interaction effects for mortality between OAG and other covariates, such as sex, diabetes, and hypertension.
Results from the multivariate-adjusted models showed a borderline significant association between higher levels of IOP and cardiovascular mortality (RR, 1.02; 95% CI, 1.00-1.04). These results also suggested that IOP-lowering treatment at baseline was related to an increased risk of cardiovascular mortality (RR, 1.42; 95% CI, 0.94-2.17). Table 4 presents results of analyses that evaluated mortality in persons with ocular hypertension and OAG, with and without previous treatment (adjusting for age, sex, hypertension, diabetes, cardiovascular disease, and stroke). The RRs showed an estimated 28% excess cardiovascular mortality for ocular hypertension compared with the group without ocular hypertension at borderline significance levels. As such, this finding was consistent with the suggested increase in cardiovascular deaths with increasing IOP (Table 3). Results did not vary appreciably when the 32 treated persons were excluded from the ocular hypertension group.
Similarly, there was an excess 38% cardiovascular mortality in the subset with previously diagnosed and treated OAG (Table 4). Therefore, we further investigated the relationship between mortality and type of IOP-lowering medications. Timolol maleate was the most frequently used IOP-lowering medication in this population (60%); it was used alone (18%) or combined with pilocarpine and/or other medications, such as acetazolamide. Table 5 presents the associations of mortality and timolol maleate treatment at baseline, both in persons with OAG only (including treated OAG [n = 141]) and in the overall population (including both treated OAG and ocular hypertension [n = 173]). In persons with OAG, any treatment at baseline tended to increase both all-cause (RR, 1.45; P = .07) and cardiovascular mortality (RR, 1.64; P = .06); these RRs were slightly higher than those estimated for the overall population (1.30 for all-cause and 1.42 for cardiovascular mortality). Among those with OAG, timolol maleate treatment (alone or combined with other treatment) significantly increased all-cause (RR, 1.70; 95% CI, 1.08-2.68) and cardiovascular (RR, 1.91; 95% CI, 1.04-3.50) mortality compared with untreated OAG. When the relationship was evaluated in the overall population, increased risks for all-cause mortality (RR, 1.43; P = .06) and cardiovascular mortality (RR, 1.58; P = .07) were also suggested. Analyses limited to participants with treated ocular hypertension were not meaningful because of the small number treated with timolol maleate at baseline in this group (n = 22) .
This report from the Barbados Eye Studies provides the first long-term data on the links between OAG and mortality in a population of African origin. After 9 years of follow-up, the increased number of deaths in persons with OAG (observed for all-cause and cardiovascular mortality) was largely explained by the advanced age of this group (Table 3). However, persons with prior OAG diagnosis and treatment tended to have increased risk of cardiovascular deaths (Table 4). In addition, persons with OAG receiving timolol maleate (alone or combined) at baseline had significantly increased risks of all-cause (70%) and cardiovascular (91%) mortality (Table 5). In this black population, increasing IOP levels were related to increased mortality from cardiovascular diseases (RR, 1 .02, per 1 mm Hg increase in IOP) (Table 3), with an estimated 28% excess mortality for persons with ocular hypertension (RR, 1.28) (Table 4).
Reports evaluating associations between survival and glaucoma have been inconsistent.14 Differences in study design, definitions, and populations may partly explain the variation in results. No significant associations between glaucoma and mortality were found in population-based studies in which the diagnosis of glaucoma was based on standardized examinations.8- 13 The most recent of these reports originates from the Blue Mountains Eye Study, which also failed to find significant associations between mortality and overall glaucoma.13 The nonsignificant relationship in all these studies appears to agree with our results. As in most of these studies,9- 12 we observed an excess mortality before, but not after, adjustment for multiple coexisting risk factors for decreased survival.
In contrast, Lee and colleagues7 reported a positive association between self-reported glaucoma and mortality (all-cause and cardiovascular disease) among adult participants in the US National Health Interview Survey (1986-1994). Hiller et al6 found that a high IOP or a history of glaucoma treatment was associated with decreased survival in the Framingham Eye Study cohort. The Wisconsin Epidemiologic Study of Diabetic Retinopathy also showed that after controlling for age and sex, history of glaucoma was associated with all-cause and ischemic heart disease mortality in a population with younger-onset diabetes.23 The positive relationships found in the latter studies were based on self-reported history of glaucoma or treatment, which could erroneously include diagnosis and treatment of ocular hypertension, a condition associated with increased mortality in our study.14
While the Blue Mountains Eye Study showed no association between mortality and overall glaucoma, it reported an increased cardiovascular mortality in 55 persons with previously diagnosed glaucoma (RR, 1.85; 95 % CI, 1.12-3.04).13 Similarly, our study found a borderline significantly increased risk of cardiovascular mortality in persons with previously diagnosed and treated OAG (RR, 1 .38) (Table 4). In addition to having standardized protocols for glaucoma classification, the Blue Mountains Eye Study and Barbados Eye Studies collected and verified data on the use of any medications, thus minimizing reporting biases. One explanation for the excess mortality found in persons with previously diagnosed OAG could be their longer duration of disease compared with those with newly diagnosed disease. However, those with newly diagnosed disease do not necessarily have a shorter disease duration because the date of OAG onset is unknown.
Another explanation for an increased mortality risk could be related to the OAG treatment received. In fact, the adverse effects of glaucoma treatment possibly explain the association between OAG and mortality found in some studies.7,24 Topical therapy with β-blocking drugs, commonly prescribed in OAG, lowers IOP by reducing the inflow of aqueous humor via β-adrenergic mechanisms in the ciliary epithelium.25,26 Survival could decrease owing to adverse effects linked to these medications (including potentially serious cardiovascular events, respiratory adverse effects, and others25- 27) or from inappropriate use (contraindications). In the Blue Mountains Eye Study, significant associations were found between cardiovascular mortality and glaucoma treated with timolol maleate.13 In the Early Manifest Glaucoma Trial, patients treated with betaxolol hydrochloride and argon laser trabeculoplasty had a higher mortality than controls after a median follow-up of 6 years, though the difference was not statistically significant (P = .08).28 The Malmö Screening Study,8 which was based on the same population as the Early Manifest Glaucoma Trial, as well as including self-selected patients with glaucoma, indicated no associations between glaucoma treatment and increased mortality. The Ocular Hypertension Treatment Study found no differences in death rates between patients with treated and with untreated ocular hypertension after a similar length of follow-up as in the Early Manifest Glaucoma Trial,29 though the Ocular Hypertension Treatment Study included younger patients without glaucoma. When we evaluated the relationship, a similar association between any treatment and all-cause mortality was suggested in the subset of persons with OAG (RR, 1.45; P = .07) (Table 5), a finding comparable with that from the Early Manifest Glaucoma Trial report. In the Barbados Eye Studies, cardiovascular mortality risk also tended to be higher for treated than untreated OAG (RR, 1.64; P = .06) (Table 5).
Results from the Barbados Eye Studies did not provide conclusive evidence about the potential role of overall IOP-lowering treatment on mortality. However, these results were consistent with data from the Blue Mountains Eye Study, in that risk of cardiovascular mortality was increased in participants with previously diagnosed glaucoma and treated with any timolol maleate eye drops at baseline.13 In the Blue Mountains Eye Study, a dose-dependent pattern of association between duration of timolol maleate use and the magnitude of cardiovascular mortality risk was also observed; data on duration of specific treatments were not available from the Barbados Eye Studies. Possibly, the association of mortality with any timolol maleate use could be a marker for longer treatment and/or more advanced glaucoma rather than the medication itself causing mortality. Although the combination of medications other than timolol maleate indicated a weaker association with mortality, as shown by the lower RRs in Table 5, we did not evaluate each medication individually because of the small numbers. It is possible that treatment involving timolol maleate was more effective and better tolerated than other medications and was thus used longer and therefore there were more observed deaths. While adverse effects and disadvantages of β-adrenergic antagonists such as timolol maleate have been reported in various studies13,25- 27,30 and could contribute to early deaths, our findings could be because of chance and are subject to the inherent limitations of the observational study design.
Evidence shows that risk factors for glaucoma may not be the same as those for ocular hypertension, suggesting different etiologic mechanisms.6,31,32 Although the unadjusted excess mortality for OAG in our study was mostly explained by aging and additional confounding factors, higher levels of IOP significantly increased the risk of cardiovascular mortality. When ocular hypertension was defined as having an IOP greater than 21 mm Hg or as treatment in persons without glaucoma, we noticed a modest excess mortality relating to cardiovascular disease (28%). These results are consistent with data from the Framingham Eye Study cohort; high IOP (≥ 25 mm Hg) or a history of glaucoma treatment tended (P = .08) to be associated with an approximately 40% decreased survival.6
Diabetes and hypertension are established risk factors for ocular hypertension but not glaucoma,31- 33 and may in part explain the findings of increased mortality. Additional analyses found no significant interaction of IOP with diabetes or hypertension regarding mortality. Although the Barbados Eye Studies data could not verify these systemic conditions as effect modifiers, uncontrolled confounding remained possible. Regression analyses that examined possible risk factors for elevated IOP often found that these conditions, eg, diabetes and hypertension, explained a very low percentage of the variation (approximately 10%) in IOP; as such, other main factors contributing to high IOP remain unknown.34- 36
While OAG has been established as an age-related ocular disease, the relationship between IOP and increasing age has been inconsistent, with some studies showing increases in IOP with age and others reporting a null or negative association between ocular pressure and aging.37,38 Therefore, ocular hypertension may be less of an age-dependent condition than glaucoma. This possibility may partly explain why accounting for age nullifies the excess mortality in persons with OAG, but not in persons with higher IOP, in whom a modest risk persists.
Ocular hypertension may share a similar pathway as that which leads to cardiovascular diseases, with the latter conditions ultimately decreasing survival. The relationship between ocular hypertension and increased mortality needs to be confirmed; reasons for their relationship should be investigated in future studies.
Major strengths of the study included its long-term follow-up and good participation, which was particularly good for a population-based study of an older cohort. Additionally, information on deaths was carefully ascertained and verified using death certificates. Classification of glaucoma was based on standardized examinations and specific criteria that were independent of IOP levels. Furthermore, participants were asked to bring all medications to their study visits for verification. Although such medications were verified for more than 85% of the participants reporting IOP-lowering treatment, reporting bias is possible among persons without verification; also, information on adherence to therapy was not available. Also, our analyses are based on treatment received at baseline and do not consider treatment received during follow-up, which may weaken the specificity of associations with timolol maleate treatment.
Variations in IOP measurements, which may peak on a circadian basis, could lead to misclassification. However, IOP for all participants was measured during office hours and no significant relationship was observed between diurnal variation and IOP.31 The IOP level of 21 mm Hg (80th percentile in the Barbados Eye Studies), used as a criterion to define ocular hypertension, was arbitrary and mainly used for categorizing IOP data for evaluations of mortality. Nonetheless, the findings of these categorical analyses were supported by the results based on IOP as a continuous variable. Blood pressure–lowering medications may reduce IOP levels and could be a source of misclassification. However, results remained very similar after additional controlling for antihypertensive treatment (data not shown). Our analyses adjusted for several potential risk factors, including hypertension and diabetes, but other concomitant disorders and their medications are likely to be relevant in this older population; as such, the possible effects of uncontrolled confounding could not be ruled out.
Although our results are based on a population with the same ancestral origin as African American individuals,39 questions may be raised as to the comparability regarding morbidity/mortality patterns and access to care. The prevalence of conditions like diabetes and hypertension is similar in African Barbadian individuals and African American individuals,40- 42 as are the patterns of all-cause and cardiovascular mortality.43 Life expectancy is high (approximately 76 years) and health care is free in Barbados, and African Barbadian citizens (being the majority population [≫ 90%]) span all socioeconomic strata. Such information may assist in interpreting our findings.
In the black population of the Barbados Eye Studies, who have the same morbidity and mortality patterns as African American individuals, the risk of mortality was not independently associated with OAG. However, persons with previously diagnosed OAG who were receiving treatment at baseline tended to have increased mortality after 9 years of follow-up. In particular, a significantly increased risk of mortality with treatment involving timolol maleate was noted in persons with OAG. These results support previous findings in a white population and warrant further investigations, ideally in randomized clinical trials. Furthermore, an estimated excess cardiovascular mortality of 28% for persons with ocular hypertension was also suggested. These findings underscore the importance of close monitoring and controlling of adequate IOP levels in this and other high-risk populations.
Group Information: A list of the Barbados Eye Studies members was published in Arch Ophthalmol. 2006;124(11):1631-1636.
Correspondence: Suh-Yuh Wu, MA, Department of Preventive Medicine, Stony Brook University, HSC L3 Room 086, Stony Brook, NY 11794-8036 (firstname.lastname@example.org).
Submitted for Publication: December 14, 2006; final revision received April 12, 2007; accepted April 21, 2007 .
Financial Disclosure: None reported.
Funding/Support: This study was supported by grants EY07625, EY07617, and EY014921 from the National Eye Institute, Bethesda, Maryland.