Prevalence of Open-Angle Glaucoma Among Adults in the United States FREE

The most recent estimates of the burden of open-angle glaucoma (OAG)in the United States relied on limited data.¹ Oneobstacle to obtaining accurate estimates is the lengthy examination proceduresneeded to identify individuals with glaucoma. Detecting glaucoma in eye diseaseprevalence surveys requires detailed evaluation of both the optic nerve headand the visual field. Fortunately, several recent major population-based surveyshave determined the prevalence of glaucoma using rigorous study designs.^{2 - 14}

The aim of this research was to use pooled data from these large, worldwidepopulation-based studies to determine more precisely the magnitude of theproblem in the United States and to project how the numbers will change inthe coming decades.

Principal investigators from the following studies provided data onthe prevalence of OAG: the Baltimore Eye Survey,² theBarbados Eye Study,⁴ the Beaver Dam Eye Study,³ the Blue Mountains Eye Study,⁵ theKongwa Eye Project,¹⁵ Proyecto Vision EvaluationResearch,⁸ the Rotterdam Study,¹⁰ andthe Melbourne Visual Impairment Project.⁶ Table 1 provides the baseline demographicsof subjects in each of the studies contributing data for the present research.The Barbados and Tanzania data were excluded from the main estimates of USprevalence, but included in alternative analyses.

The Baltimore Eye Survey (1985-1988) enrolled 5308 black and white subjects(75% of the intended population); the Beaver Dam Eye Study (1988-1990) inBeaver Dam, Wis, enrolled 4926 subjects (83% of the intended population);the Blue Mountains Eye Study (1992-1994) in Sydney, Australia, examined 3654white subjects (82% of the eligible population); the Rotterdam Study (1990-1993)enrolled 6774 subjects (67% of the intended population); Proyecto Vision EvaluationResearch enrolled 4774 subjects (72% of the eligible population); and theMelbourne Visual Impairment Project (1991-1998) enrolled 4744 persons (86%response rate). The investigators from each of those studies provided us withthe number of individuals able to undergo evaluation for OAG and the numberwith definite OAG stratified by gender and race for groups aged 40 to 44,45 to 49, 50 to 54, 55 to 59, 60 to 64, 65 to 69, 70 to 74, 75 to 79, and80 to 84 years and 85 years or older.

There is no single standard for defining OAG in population-based research.Researchers have instead relied on a wide range of approaches, including consensusmeetings,⁶ review of all suspected cases bya single expert,² and statistical approachesusing cutoffs for cup-disc ratio and visual field defects to define the disease.^{10 ,12} For the purposes of this research,studies were eligible to contribute data if the determination of glaucomawas made using both visual field and photographically obtained optic nervehead data. The definitions used in the included studies are presented in Table 2.

To be included, studies had to contribute data believed to be directlyapplicable to the US population. Some recent publications from populationsoutside the United States were not included because it is not clear that thefindings from those populations are representative of what one would expectin those minority populations who have emigrated to the United States.^{4 ,9 ,12 ,15 - 16} Furthermore,we were unable to obtain data from some studies meeting inclusion criteriawithin the time allocated to this project.^{7 ,11}

The age-specific prevalence proportions were derived in 2 steps. First,pooled prevalence proportions were estimated for each race-, gender-, andage-specific stratum using minimum variance linear estimation. Stratum-specificproportions from each study were transformed using a logarithm odds transformation,and proportion variances were based on the binomial distribution. Second,logistic regression models were fitted to the pooled prevalence proportionsusing the midpoint of each age interval as the independent variable. Modelswere fit separately by race and gender. Prevalence estimates for black andHispanic persons were based on modeled rates from a single study. No prevalencedata were available for other minority US populations; therefore, estimatesfor other races were based on modeled rates using the unweighted average ofthe pooled stratum-specific rates for white, black, and Hispanic persons.

Age and race effects in the models were evaluated using logistic regressionand the Wald χ² test statistic. Odds ratios for gender differenceswere based on Mantel-Haenszel χ² tests for the 2 × 2tables of observed rates, adjusting for age and the study effect.

The number of people with OAG in the United States in each race, gender,and age category were generated by applying the modeled prevalence rate foreach year of age to the 2000 US census population and summing across the agerange for each 5-year age category. Projected estimates were derived in thesame manner, using US Census middle series population projections for theyear 2020. Stratum-specific US prevalence rates were computed by dividingthe total number of estimated cases for each stratum by the stratum-specificUS population. Estimates for glaucoma in Western Europe and Australia werebased on applying the gender- and age-specific rates for white persons totheir respective populations 40 years and older.

The age-specific prevalence of OAG among white, black, and Hispanicpersons from each of the studies is presented in Figure 1. Focusing separately on each of the race/ethnic groups,we found the following results.

Pooled data for European-derived individuals from the Baltimore EyeSurvey, the Blue Mountains Eye Study, the Beaver Dam Eye Study, the RotterdamStudy, and the Melbourne Visual Impairment Project found a strong increasein the prevalence of OAG with age (P <.001, χ² test). In the 50- to 54-year age range, 0.89% of white women had OAGcompared with 2.16% of those in the 70- to 74-year age range and 6.94% ofthose 80 years and older (Table 3).After controlling for age, there were no significant differences by gender(odds ratio [OR] for women, 1.03; 95% confidence interval [CI], 0.83-1.27)in the prevalence of OAG in white subjects. The estimated US prevalence ofOAG among white individuals 40 years and older is 1.69% (95% CI, 1.53%-1.85%).

Data for black subjects were derived from a single study, the BaltimoreEye Survey. The prevalence of OAG increased with age, and OAG was consistentlymore prevalent than in white subjects (Table 3). Black women aged 50 to 54 years had a prevalence of OAGof 2.24%, which increased to 5.89% for those aged 70 to 74 years, and to 9.82%for those 80 years and older. The age-adjusted prevalence of OAG was lowerin women compared with men, but did not differ significantly (OR, 0.83; 95%CI, 0.55-1.25). Logistic regression including age, race, and gender in themodel found that black subjects had almost 3 times the prevalence of OAG comparedwith white subjects (OR, 2.82; 95% CI, 2.14-3.72).

The data on Hispanic subjects came from a single study of mostly Mexican-derivedLatinos from Arizona.⁸ Prevalence estimatesshowed similar increases with age, but with a markedly higher prevalence inthe oldest Hispanic subjects (Table 3).After controlling for age and gender, rates of OAG in Hispanic subjects didnot differ significantly from that among white subjects (OR, 1.06; 95% CI,0.89-1.26), except for those older than 65 years, in whom the rates were higher(OR, 1.24; 95% CI, 1.10-1.41). After controlling for age and gender, Hispanicsubjects had a significantly lower prevalence of glaucoma than black subjects(OR, 0.41; 95% CI, 0.27-0.60). Although women had somewhat higher age-adjustedrates of OAG than men, the difference was not statistically significant (OR,1.11; 95% CI, 0.72-1.71).

The overall prevalence of OAG in the US population 40 years and olderis estimated to be 1.86% (95% CI, 1.75%-1.96%), with 1.57 million white and398 000 black persons affected (Table4). Applying race-, age-, and gender-specific rates to the US populationas determined in the 2000 US census, we estimate that OAG affects 2.22 millionUS citizens. Owing to the rapidly aging population, the number with OAG willincrease by 50% to 3.36 million in 2020.

Applying the same age-, race-, and gender-specific rates, the numberof affected individuals with OAG is estimated at 122 000 in Australia,and 3 million in Western Europe.

Pooled data from population-based eye disease prevalence studies indicatethat, at present, 2.22 million individuals in the United States have open-angleglaucoma. This estimate is similar to the one made almost a decade ago byTielsch¹ for the 1990 United States population(2.0 million). Studies consistently find that about half of those with glaucomaare unaware they have the disease.^{2 ,4 - 6} Recentreports indicate that lowering intraocular pressure prevents vision loss inpatients with glaucoma and ocular hypertension,^{17 - 19} somost of those individuals with undiagnosed OAG could potentially benefit fromtreatment. Furthermore, many more are eligible for care, since we did notestimate the prevalence of ocular hypertension without signs of glaucoma.We project that the number of individuals with potentially treatable OAG willincrease from 2.22 million today to more than 3 million in 2020. This hasimplications for the health care system in the United States. Glaucoma wasthe fifth most common diagnosis for an office visit by Medicare recipientsin 1992.²⁰ Western Europe has an even higherprevalence of OAG, largely because of the older age structure there.

These numbers are particularly concerning because glaucoma leads toirreversible vision loss. Recent research indicates that those with mild-to-moderateglaucomatous visual field loss have decreased mobility,²¹ andthose with visual field loss due to any cause are more likely to report falling.²² Those with more severe forms of the disease are oftenhighly dependent on others. In addition, glaucoma management is expensiveand not without risk. Medications can lead to breathing and cardiac problems,^{23 - 26} andsurgery to lower eye pressure is associated with ocular discomfort,²⁷ cataract formation,^{28 - 29} andendophthalmitis.^{30 - 31}

The present research has several limitations. First, although this isa meta-analysis of population-based studies, none of the studies enrolledall eligible subjects. On average, about 20% of those eligible did not participate,which may cause bias in the estimates. Nonparticipants may include more individualswith known disease, as these persons may not see any benefit to participating.Conversely, nonparticipants may have had better ocular health and did notparticipate because they saw no value in a free eye examination. Furthermore,to diagnose glaucoma, most studies relied on visual field and optic nervehead data and results of a definitive eye examination. Some did not attendthe final eye examination and were therefore excluded from a diagnosis. Ifthese individuals were more likely to have glaucoma than those who attendedthe examinations, then estimates may be lower than the true prevalence.

A second limitation is the lack of a gold standard for diagnosing glaucomain prevalence surveys. Each investigative team used its own approach to definethe disease. However, even with this variation in methods, the results wereremarkably similar across studies, indicating that researchers were capturingthe same condition (on average) or that rates were actually more variablebut that variation was missed owing to the different definitions. In eithercase, we assumed that if both disc and visual field data were used in definingglaucoma without regard to intraocular pressure, then the definition was likelyto be accurate.

A third limitation is the relatively sparse data on black and Hispanicsubjects. We elected to exclude data from well-designed studies in black populationsfrom outside the United States. Prevalence rates from Barbados and St Lucia,⁹ both Caribbean populations originating in West Africa,were substantially higher than those found in a US black population, and weretherefore excluded. If these populations more accurately reflect the trueprevalence of glaucoma in the United States than the Baltimore Eye Study data,then we would have underestimated the prevalence of OAG among black subjectsin the present report. The prevalence data from Tanzania, although similarto that found in the United States, were also not included because most AfricanAmericans who are descendents of the slaves trace their origins to West Africa,an area with different ethnic groups from East Africa.³² Inaddition, the study population was derived from a single ethnic group fromthis region.

To assess the possible underestimation that resulted from excludingthose studies, Table 5 shows theprevalence of OAG among black subjects and the number affected in each age-,race-, and gender-specific category, using data from Baltimore alone, datafrom Barbados alone, and combined data from Baltimore, Barbados, and Tanzania.Although we believe that the most applicable estimate for the United Statescomes from the Baltimore data, had we pooled the data from Barbados and Tanzaniawith those from Baltimore, the estimated number of affected black personsin the United States in 2000 would be 583 000 (a prevalence of 4.9% asopposed to our estimate of 3.4%). For Hispanic persons, all estimates werebased on a single study of a select population from Arizona. These resultsmay be different from those that would be found if other Hispanic populationswere studied.

A final important limitation is the lack of data on other minority USpopulations. Given the total absence of data on these US populations, we estimatedthe rates for this group on the basis of an unweighted average of the ratesfound for black, white, and Hispanic subjects. These estimates will thereforehave to be revised as more data are collected in these populations. Otherrecent studies from Asia have findings that may be relevant to US populations.We have chosen not to include data from Chinese populations in Singapore andelsewhere, as US census data do not clearly distinguish among the differentAsian populations.

This report gives the best available estimate for the magnitude of theproblem of OAG in the United States based on a meta-analysis of population-baseddata. The number of US population affected by OAG is large, including morethan 2 million people at present, and the aging population will increase thissubstantially in the years to come. Previous work indicates that more thanhalf of these individuals are unaware that they have the disease and willlikely suffer unnecessary vision loss. Better detection and effective, safe,and early interventions are needed to minimize the impact that glaucoma willhave on our aging population.

Corresponding author: David S. Friedman, MD, MPH, Wilmer Eye Institute,The Johns Hopkins Hospital, 600 N Wolfe St, Baltimore, MD 21287(e-mail: david.friedman@jhu.edu).

Submitted for publication May 25, 2003; final revision received November24, 2003; accepted December 8, 2003.

From the Data Center for Preventive Ophthalmology, Wilmer Eye Institute,The Johns Hopkins University, Baltimore, Md (Drs Friedman, West, Congdon,Kempen, and Tielsch); the Department of Epidemiology and Biostatistics, Schoolof Public Health and Health Services, George Washington University MedicalCenter, Washington, DC (Ms O'Colmain); Macro International, Inc, Calverton,Md (Ms O'Colmain); the Department of Ophthalmology, Erasmus Medical Center,Rotterdam, the Netherlands (Dr Wolfs); the Department of Ophthalmology, Universityof Wisconsin, Madison (Dr Klein); the Centre for Eye Research Australia, Universityof Melbourne, East Melbourne, Victoria, Australia (Dr Taylor); the Departmentof Preventive Medicine, School of Medicine, Stony Brook University, StonyBrook, NY (Dr Leske); and the Department of Ophthalmology, University of SydneyEye Clinic, Westmead Hospital, Sydney, New South Wales, Australia (Dr Mitchell).

The members of the Eye Diseases Prevalence Research Group are as follows:

The Baltimore Eye Survey, Baltimore, Md: JamesM. Tielsch, Alfred Sommer, Joanne Katz, and Harry A. Quigley. The Barbados Eye Study, Barbados, West Indies: M. Cristina Leske, Suh-YuhWu, Barbara Nemesure, Anselm Hennis, Leslie Hyman, and Andrew Schachat. The Beaver Dam Eye Study, Beaver Dam, Wis: Scot Moss, BarbaraE. Klein, Ronald Klein, Kristine E. Lee, and Sandra C. Tomany. Blue Mountains Eye Study, Sydney, New South Wales, Australia: PaulMitchell, Jie Jin Wang, Elena Rochtchina, Wayne Smith, Robert G. Cumming,Karin Attebo, Jai Panchapakesan, Suriya Foran. The MelbourneVisual Impairment Project, Melbourne, Victoria, Australia: Hugh R.Taylor, Cathy McCarty, Bickol Mukesh, LeAnn M. Weih, Patricia M. Livingston,Mylan Van Newkirk, Cara L. Fu, Peter Dimitrov, Matthew Wensor. Proyecto VER (Vision Evaluation Research), Nogales and Tucson, Ariz:Sheila West, Beatriz Muñoz, Jorge Rodriguez (deceased), Aimee Broman,Daniel Finklestein, Robert Snyder. Rotterdam Eye Study,Rotterdam, the Netherlands: Paulus T. V. M. de Jong, M. Kamran Ikram,Caroline C. W. Klaver, Roger C. W. Wolfs, Simone de Voogd, Johannes Vingerling,Redmer van Leeuwen. Salisbury Eye Evaluation Project, Salisbury,Md: Sheila West, Gary Rubin, Karen Bandeen Roche, Beatriz Muñoz,Kathy Turano, Oliver Schein, Donald Duncan.