Causes and Prevalence of Visual Impairment Among Adults in the UnitedStates FREE

Blindness and low vision are widely recognized as important causes ofimpairment among Americans.^1- 3 Becauseof the cost and logistical difficulty of carrying out an appropriate samplingscheme, few population-based studies of national scope have been carried outin the United States to estimate the prevalence of visual impairment. WhileThe National Health and Nutrition Examination Survey (NHANES) has collectednational data on vision disabilities,⁴ thelack of photographic documentation of ocular conditions limits the abilityto assess causal associations. In addition, these data were collected morethan 30 years ago. The Visual Acuity Impairment Survey carried out by theNational Eye Institute in the 1980s suffered similar limitations because ofthe absence of photographic documentation and the large proportion of personswho could not be examined.⁵ Population-basedstudies of ocular disease that have been carried out in the United States^6- 9 are potentiallylimited in their generalizability by local variations in the populations studied,accessibility of eye care, and patterns of surgical practice.

Recognizing the need for national estimates of visual impairment, PreventBlindness America (Schaumburg, Ill) and the National Eye Institute (Bethesda,Md) invited the principal investigators of several population-based visionstudies to a meeting in Fort Lauderdale, Fla, in May 2001, to standardizedisease definitions and methods of data reporting so that available data frommany of these studies might be analyzed together. Age- and race/ethnicity-specificprevalences of blindness and low vision were calculated based on studies ofbest-corrected visual acuity from the United States; Western Europe; Barbados,West Indies; and Australia. These estimates were then applied to the populationstructure of the United States as reported in the 2000 census,¹⁰ toestimate the number of visually impaired persons nationally. Projections ofprevalence and number of visually impaired persons in 2020 were also madebased on census projections for the US population in that year. These resultsrepresent the most robust and up-to-date estimates available of the burdenof visual impairment facing the United States.

A decision was made in principle to include all population-based studiesof blindness and low vision conducted in North America, Western Europe, andAustralia and published in English after 1990 and up to the start of the currentproject (May 2001) (Table 1 and Figure 1). The cutoff year of 1990 was chosenboth owing to the scarcity of scientifically valid studies prior to this timeand to minimize potential inaccuracies caused by secular trends in treatmentand surgical practice over time. While studies from Western Europe and Australiawere included in estimates for European-derived persons, potentially relevantstudies from Africa¹¹ and Barbados¹² were excluded from prevalence estimates for African-derivedpersons because of concerns over the potential effect of rates of medicaland surgical treatment being significantly different from those in the UnitedStates. Data from Barbados were evaluated; but since the prevalence was substantiallyhigher than in studies of black persons in the United States, these data wereexcluded from the estimation of prevalence; however, they were used in theestimation of causes of blindness among black persons.

Principal investigators from studies listed in Table 1 were asked to furnish data tables listing the prevalenceof blindness and low vision in the better-seeing eye by 5-year age interval,gender, and (where relevant) race/ethnicity. Data were requested for boththe World Health Organization (WHO) (<6/120 [<20/400]) and US (≤6/60[≤20/200]) definitions of blindness, while low vision was defined as worsethan 6/12 (<20/40) in the better-seeing eye, excluding those who were categorizedas being blind by US definition. All measurements related to best-correctedvisual acuity. Where not otherwise specified, all estimates for blindnessare reported using the US definition, 6/60 (20/200) or worse in the better-seeingeye.

Investigators were asked to attribute a cause to the blindness and lowvision for all persons with bilateral blindness or low vision. However, nospecific causes of low vision were attributed by either the Beaver Dam EyeStudy, Beaver Dam, Wis, or the Melbourne Visual Impairment Project, Melbourne,Victoria, Australia. If it was felt that more than one disease entity wasresponsible for visual impairment in a subject, then a subjective decisionwas made in assigning the primary cause. This was possible for all subjects,with the exception of 2 participants in the Blue Mountains Eye Study, Sydney,New South Wales, Australia, both of whom received the diagnoses of cataractand glaucoma.

The age-specific prevalence proportions for blindness and low visionwere derived by pooling the race/ethnicity- and age-specific prevalences fromall contributing studies using minimum variance linear estimation. Stratum-specificproportions from each study were transformed using a logarithm odds transformationand proportion variances were calculated assuming a binomial distribution.The low numbers of cases in some 5-year age and gender intervals were insufficientto provide robust, pooled, gender-specific estimates, so these estimates werecollapsed over gender. To determine the effects of age, sex, and race/ethnicityon the prevalence of disease and to produce a smoothed function over the ageintervals for estimating the prevalence of the disease, logistic regressionmodels were fit to the pooled prevalence proportions using the midpoint ofeach age interval as the independent variable. Models were fit separatelyfor black, white, and Hispanic persons. For purposes of calculating the overallUS prevalence, the estimates for other racial/ethnic groups were modeled usingthe average of the pooled age-specific estimates for these 3 groups.

The number of visually impaired persons in the United States in eachrace/ethnicity and age category was generated by applying the modeled prevalencefor each year of age to the 2000 US census population and summing across theage range for each 5-year age category. Projected estimates were derived applyingthe modeled prevalence for each year of age by race/ethnicity to the 2000US census middle-series projections for 2020. Constant age- and race/ethnicity-specificprevalences were assumed over this period for both blindness and low vision.Stratum-specific US prevalences were computed by dividing the total numberof estimated cases for each stratum by the stratum-specific US population.

Age and race/ethnicity effects were tested using the model Wald χ² test statistics. Odds ratios for race/ethnicity were derived fromlogistic regression coefficients for the appropriate racial comparisons. Testsfor gender differences were based on the observed age-, race/ethnicity-, andgender-specific prevalence from each study. Separate Mantel-Haenszel χ² tests for 2 × 2 tables of observed rates were carried out byrace/ethnicity, controlling for both age and study effects.

In 2000, there were an estimated 937 000 blindAmericans who were older than 40 years (US definition), a prevalenceof 0.78% (95% confidence interval [CI], 0.63%-0.94%). The number of personswith low vision was estimated to be an additional 2.4 million (1.98%; 95%CI, 1.74%-2.21%), for a total of 3.3 million Americans aged 40 years and olderwith visual impairment (Table 2).

The leading cause of blindness among white Americansin 2000 was age-related macular degeneration (AMD), accounting for54% of all blindness, as opposed to 9% for cataract, the next most commoncause (Figure 2). Age-related maculardegeneration was also the leading cause of blindness in all 6 population-basedstudies of European-derived populations on which the current article is based.The leading causes of blindness among black persons were more evenly distributedthan among white persons, with cataract and open-angle glaucoma (OAG) accountingfor slightly over 60% of the cases. Among Hispanic persons,OAG was the most common cause of blindness (28.6% of the cases),although this is based on a total of only 4 persons bilaterally blind fromglaucoma.

Cataract was the most frequently reported condition in persons withlow vision, associated with approximately 50% of low vision cases among black,white, and Hispanic persons. This was more than twice the proportion of anyother condition (Figure 3).

Both low vision and blindness increased significantly with age for allraces/ethnicities (P <.001 for all races/ethnicitiesfor both blindness and low vision) (Table3). Although limited data did not permit us to make robust estimatesin the age range beyond 85 years, a highly significant quadratic term in ourmodels, together with a rapid apparent increase in estimated blindness prevalenceabove the age of 85 years, suggests that blindness prevalence rises rapidlyamong the oldest segment of the population.

Age-specific blindness prevalence was higher for black persons comparedwith white persons (odds ratio [OR] = 2.77; 95% CI, 1.56-4.92) or Hispanicpersons (OR = 3.13; 95% CI, 2.29-4.29), while the prevalence of low visionamong Hispanic persons was higher than that for white persons OR = 1.39; 95%CI, 1.24-1.56) and black persons (OR = 1.90; 95% CI, 1.28-2.83) (Table 3).

Men had a significantly higher age-adjusted prevalence of blindnessthan women among black persons (P = .002), but notamong white (P = .20) or Hispanic persons (P = .64). The age-adjusted prevalence of low vision wassignificantly higher for women among white persons (P =.01) but did not differ significantly by gender among black (P = .96) or Hispanic persons (P = .11). (Table 4).

It was estimated that the number of blind persons olderthan 40 years in the United States would increase by approximately 70% to1.6 million (prevalence of 1.1%) by 2020. The number of persons with low visionwas projected to be 3.9 million (2.5%), for a total of 5.5 million visuallyimpaired Americans (3.6%).

Estimates of the prevalence of visual impairment among Australians olderthan 40 years in 2000 were as follows: low vision 143 000 (1.8%), blindness(US definition of ≤6/60 [≤20/200] best-corrected visual acuity in thebetter-seeing eye) 55 000 (0.7%), and blindness (WHO standard of <6/120[<20/400] best-corrected visual acuity in the better-seeing eye) 35 000(0.4%). The corresponding figures for Western Europe were low vision, 3.64million (2.0%); blindness (US definition); 1.38 million (0.7%); and blindness(WHO standard), 879 000 (0.5%). These estimates were derived by applyingthe age-specific modeled prevalence rates for white persons to the populationsof Australia and Western Europe in each 5-year age interval.

Visual impairment is highly prevalent in the United States and is projectedto increase rapidly as the population ages. Our results were consistent withthose of the participating studies in demonstrating that the prevalence ofvisual impairment increases with increasing age. Diseasesof aging are also the most common causes of blindness (AMD, cataract, andglaucoma) and of low vision (cataract).

Our finding of a higher age-adjusted blindness prevalence among blackmen as opposed to black women runs contrary to the results of a recent meta-analyticstudy that reported a significant age-adjusted excess of blindness among womenin population-based studies throughout Africa, Asia, and the developed world.¹³ Overall, blindness was significantly (P = .004) more common in white women than white men, but the differencewas no longer significant after adjustment for age (P >.10).The lack of an increased burden of blindness among white women is counterto a recent report from a large, older, white population in England that womenare at greater age-adjusted risk for blindness than men.¹⁴ Whitewomen did appear to be at an increased risk of low vision in the current study.

Our decision not to include population studies of blindness carriedout prior to 1990 has led to the exclusion of some important data, most notablyfrom the Framingham Eye Study.¹⁵ As noted earlier,it was felt that differing methods and the likely influence of secular trendsin diagnosis and treatment of eye disease, such as the widespread use of panretinalphotocoagulation and automated perimetry, render these data of questionableusefulness for estimates of the causes and prevalence of visual impairmentin 2000.

On the other hand, the decision to include data from Europe¹⁶ and Australia^17- 18 incalculating our estimates may equally well be questioned in the light of potentialdifferences in access to eye care and treatment patterns between these countriesand the United States. It is reassuring to note the high degree of similarityin age-specific blindness prevalence among the 6 studies of European-derivedindividuals on which our study estimates are based (Figure 1A). Differences in the observed age-specific prevalenceof low vision among studies are more prominent. However, the Baltimore EyeSurvey, Baltimore, Md, is at the lowest end of this range, and the BeaverDam Eye Study toward the highest end (Figure1C); these are both American studies.

Our estimates of the number of Americans with visual impairment areslightly lower than those given in NHANES I-A (NHANES Augmentation Study I),which reported that 7.7% of African American and 4.1% of non-Hispanic whitesubjects had binocular visual acuities of 6/15 (20/50) or worse.¹⁹ However,the NHANES I-A results are for presenting rather than best-corrected visualacuity as in our analysis, and they included only persons aged 25 to 74 years.The proportion of persons older than 65 years with blindness and low visionin the United States seems to be somewhat lower than that reported among apopulation-based sample of similar-aged British persons.²⁰ Prevalenceof visual impairment in the British sample (WHO low vision standard <6/18[20/60]) was 3.1% for persons aged 65 through 74 years, as opposed to 1.47%for Americans aged 65 through 69 years, and 2.61% for those aged 70 through74 years. Once again, however, the British study reported presenting as opposedto best-corrected visual acuity. Few national estimates exist for the prevalenceof impaired best-corrected visual acuity, as was used in the current study,presumably because of the difficulty and expense involved.

Age-related macular degeneration has been reported to be the leadingcause of blindness in most recent population-based studies of persons of Europeandescent.^{6- 8,15- 17} Thesubstantially different distribution of cause-specific blindness among African-derivedpopulations is likely the result of a combination of different factors. Thesemay include the comparatively low prevalence of neovascular AMD^21- 22 andhigh prevalence of OAG²³ among black comparedwith white persons, and poorer access to and use of eye care among black personsleading to increased visual impairment due to cataract²⁴ andOAG.²⁵ The risk for development of diabeticretinopathy also seems to be higher among black persons who have diabetesmellitus.^26- 27 This may be becauseof a combination of factors, including higher prevalence of hyperglycemiaand hypertension, poor access to medical care, and reduced health literacyamong black compared with white persons.²⁸ Thehigher burden of cataract, OAG, and diabetic retinopathy may also explainthe higher age-specific prevalence of blindness among black compared whitepersons observed in our estimates.

Although our study gives estimates for the prevalence of visual impairmentamong Hispanic persons, these data are based on a single study, Proyecto VER(Vision Evaluation Research), based in Nogales and Tucson, Ariz.⁹ Theseestimates may be susceptible to local patterns of medical and surgical practice,which are known to vary greatly by region²⁹ andpatient insurance type.³⁰ Furthermore, theestimates for overall and cause-specific visual impairment in Hispanic personsare based on a few cases. This is most apparent for blindness estimates; OAGis indicated as the leading cause of blindness among Hispanic persons in ourestimates (Figure 1) on the basisof 4 persons blind from glaucoma, compared with 2 each from diabetic retinopathy,cataract, and AMD in the Proyecto VER study.⁹ Estimatesof low vision among Hispanic persons are somewhat more robust; cataract isidentified as the leading cause of low vision among Hispanic persons in ourestimates, based on 43 persons with cataract from a total of 92 with visualacuity worse than 6/12 (<20/40).⁹

The projection of a large increase in the prevalence of blindness andlow vision in the United States during the next 2 decades is driven largelyby the fact that the prevalence of visual impairment increases sharply inpersons older than 65 years. Persons aged 80 years and older made up only7.7% of the population in our study but accounted for 69% of observed blindness.It is this group, the very old, who represent the fastest-growing segmentof the US population. To avoid this increase in visual impairment will requirebroader dissemination of prevention strategies, such as regular eye examinationsfor persons with diabetes mellitus or those with a family history of OAG,smoking cessation to reduce risk of nuclear cataract,³¹ appropriatevitamin and zinc supplements for selected persons at risk for vision-threateningAMD,³² enhanced glycemic³³ andblood pressure control among persons with diabetes mellitus, and early detectionand laser treatments for high-risk proliferative diabetic retinopathy³⁴ and clinically significant diabetic macular edema.³⁵ Further research is also needed to devise new andbetter preventive measures.

This study has important limitations. Our estimates for white personsrely in part on data from Australia and Europe, where clinical and surgicalpractices and access to care may differ from the United States. These factorsmight be expected to influence the prevalence and cause-specific distributionof visual impairment. In the case of Hispanic persons, an important and rapidlygrowing segment of the US population, estimates are based on data from a singlestudy of persons of Mexican descent,⁹ whichare likely to be influenced by local factors such as number, distribution,and practice patterns of ophthalmologists. National estimates derived in thisway may not be representative of the American Hispanic population as a whole,which comprises Cuban Americans, Puerto Ricans, and Central Americans in additionto Mexican Americans. Estimates for black persons are based on data from only2 studies,^7- 8 both from the samestate, and may be similarly unstable. It is also unlikely that estimates derivedfrom any existing studies will be fully relevant to important segments ofthe US population, such as the rural South and other ethnic minority groupssuch as Asians, Native Americans, and Native Hawaiians and other Pacific Islanders.

Diagnostic criteria for the various causes of blindness and low visionalso differed among studies. Finally, and perhaps most importantly, the estimatesreported here are based on best-corrected visual acuity. These figures donot reflect the burden of low vision and blindness due to uncorrected refractiveerror, potentially an important cause of visual impairment in the United States.^36- 37 A recent report from the MelbourneVisual Impairment Project¹⁷ suggests that almost60% of visual impairment may be due to uncorrected refractive error. If suchestimates may accurately be applied in the United States, the true numberof visually impaired Americans might be as high as 8 million. Most of thisblindness could be reversed with treatment or avoided by preventive efforts.

Nevertheless, these estimates of low vision and blindness prevalenceamong the US population are the first to take full advantage of the largenumber of population-based studies of vision carried out in the last decade.The estimates were derived from raw data provided directly by the study groupsand were combined only after agreement had been reached on standard presentationof data. A recent review has cited differences in the definition of low visionand the age range of the oldest age category as the 2 most important sourcesof disagreement in estimates of visual impairment between studies.³⁸ Our approach has allowed such differences to be eliminated.Thus, the results of the current study provide useful estimates of the burdenof visual impairment in the United States, its distribution by age, race/ethnicity,and sex, and its likely rapid future increase.

Corresponding author: Nathan Congdon, MD, MPH, Wilmer Eye Institute,Wilmer 120, 600 N Wolfe St, Baltimore, MD 21287 (e-mail: ncongdon@jhmi.edu).

Submitted for publication April 3, 2003; final revision received November19, 2003; accepted November 19, 2003.

From the Dana Center for Preventive Ophthalmology, Wilmer Eye Institute,The Johns Hopkins University, Baltimore, Md (Drs Congdon, Friedman, and Kempen,and Ms Muñoz); Macro International, Inc, Calverton, Md (Ms O'Colmain);Department of Ophthalmology, Erasmus Medical Center, Rotterdam, the Netherlands(Dr Klaver); Department of Ophthalmology, University of Wisconsin, Madison(Dr Klein); Centre for Eye Research Australia, University of Melbourne, EastMelbourne, Victoria (Dr Taylor); Department of Ophthalmology, Centre for VisionResearch, Westmead Hospital, Westmead, New South Wales, Australia (Dr Mitchell);Department of Ophthalmology, University of Sydney, Sydney, New South Wales,Australia (Dr Mitchell); and the Department of Preventive Medicine, StateUniversity of New York, Stony Brook (Dr Hyman).

This study was supported by funding from Prevent Blindness America,Schaumburg, Ill, and the National Eye Institute, Bethesda, Md.

The members of the Eye Diseases Prevalence Research Group are as follows: The Baltimore Eye Survey, Baltimore, Md: James M. Tielsch;Alfred Sommer; Joanne Katz; Harry A. Quigley. The BarbadosEye Studies, Barbados, West Indies: M. Cristina Leske; Suh-Yuh Wu;Barbara Nemesure; Anselim Hennis; Leslie Hyman; Andrew Schachat. Beaver Dam Eye Study, Beaver Dam, Wis: Barbara E. K. Klein; RonaldKlein; Kristine E. Lee; Scot E. Moss; Sandra C. Tomany. Blue Mountains Eye Study, Sydney, New South Wales, Australia: PaulMitchell; Jie Jin Wang; Elena Rochtchina; Wayne Smith; Robert G. Cumming. The Melbourne Visual Impairment Project, Melbourne, Victoria,Australia: Hugh R. Taylor; Cathy McCarty; Bickol Mukesh. The Center for Eye Research, Melbourne: 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; Jorge Rodriguez (deceased); Aimee Broman; RobertSnyder. Rotterdam Eye Study, Rotterdam, the Netherlands: Paulus T. V. M. de Jong; M. Kamran Ikram; Caroline C. W. Klaver; RogerC. W. Wolfs; Simone de Voogd; Johannes Vingerling; Redmer van Leeuwen, MD. Salisbury Eye Evaluation Project, Salisbury, Md: SheilaWest; Gary Rubin; Karen Bandeen Roche; Beatriz Muñoz; KathyTurano; Oliver Schein; Donald Duncan.