Articles

Causes of vision loss worldwide, 1990–2010: a systematic

analysis
Rupert R A Bourne*, Gretchen A Stevens*, Richard A White, Jennifer L Smith, Seth R Flaxman, Holly Price, Jost B Jonas, Jill Keeffe†, Janet Leasher†,
Kovin Naidoo†, Konrad Pesudovs†, Serge Resnikoff†, Hugh R Taylor†, on behalf of the Vision Loss Expert Group

Summary
Background Data on causes of vision impairment and blindness are important for development of public health Lancet Glob Health 2013;
policies, but comprehensive analysis of change in prevalence over time is lacking. 1: e339–49
Published Online
November 11, 2013
Methods We did a systematic analysis of published and unpublished data on the causes of blindness (visual acuity in
http://dx.doi.org/10.1016/
the better eye less than 3/60) and moderate and severe vision impairment ([MSVI] visual acuity in the better eye less S2214-109X(13)70113-X
than 6/18 but at least 3/60) from 1980 to 2012. We estimated the proportions of overall vision impairment attributable See Comment page e311
to cataract, glaucoma, macular degeneration, diabetic retinopathy, trachoma, and uncorrected refractive error in Copyright © Bourne et al. Open
1990–2010 by age, geographical region, and year. Access article distributed under
the terms of CC BY
Findings In 2010, 65% (95% uncertainty interval [UI] 61–68) of 32·4 million blind people and 76% (73–79) of *RRAB and GAS contributed
191 million people with MSVI worldwide had a preventable or treatable cause, compared with 68% (95% UI 65–70) of equally to writing of the paper

31·8 million and 80% (78–83) of 172 million in 1990. Leading causes worldwide in 1990 and 2010 for blindness were †JK, JL, KN, KP, SR, and HRT
contributed equally to the
cataract (39% and 33%, respectively), uncorrected refractive error (20% and 21%), and macular degeneration (5% and
research and paper and are listed
7%), and for MSVI were uncorrected refractive error (51% and 53%), cataract (26% and 18%), and macular in alphabetical order
degeneration (2% and 3%). Causes of blindness varied substantially by region. Worldwide and in all regions more Vision and Eye Research Unit,
women than men were blind or had MSVI due to cataract and macular degeneration. Postgraduate Medical Institute,
Anglia Ruskin University,
Cambridge, UK
Interpretation The differences and temporal changes we found in causes of blindness and MSVI have implications
(Prof R R A Bourne MD,
for planning and resource allocation in eye care. H Price PhD); Department of
Health Statistics
Funding Bill & Melinda Gates Foundation, Fight for Sight, Fred Hollows Foundation, and Brien Holden Vision Institute. and Information Systems, WHO,
Geneva, Switzerland
(G A Stevens DSc); Department
Introduction estimated that 38 million people were blind, mainly due of Genes and Environment,
Data on the causes of vision impairment and blindness to cataract, trachoma, and glaucoma. A lack of prevalence Division of Epidemiology,
form an important basis for recommendations in public data for diabetic retinopathy and macular degeneration Norwegian Institute of Public
Health, Oslo, Norway
health policies, such as planning of national budgets and precluded an estimate of their burden at that time. The
(R A White PhD); Faculty of
health services, and are important for scientific research. Global Burden of Disease, Risk Factors and Injury Study Infectious and Tropical Diseases,
Population-based studies done in the past 20 years have 2010 (GBD), which started in 2007, aimed to calculate London School of Hygiene and
revealed that cataract, glaucoma, macular degeneration, comparable estimates of burden of disease, injuries, and Tropical Medicine, London, UK
(J L Smith MSc); School of
diabetic retinopathy, trachoma, and uncorrected risk factors from 1990 and 2010.9,10 The Vision Loss Expert Computer Science and Heinz
refractive error are the most common causes of blindness Group (VLEG) of the GBD published its methods for a College, Carnegie Mellon
and vision impairment worldwide.1–4 Most studies have systematic review of published and unpublished data University, Pittsburgh, PA, USA
focused on one population within a circumscribed region from population-based studies that reported the (S R Flaxman BA); Department of
Ophthalmology,
and frequently within one ethnic group. Important prevalence of blindness and vision impairment in Universitätsmedizin,
differences between ethnic groups, regions of habitation, 1980–2012.11 We have found an decrease in age- Mannheim, Medical Faculty
demographic parameters, lifestyle, exposure to standardised prevalence of blindness and MSVI over the Mannheim, Heidelberg
environmental factors, and other factors, however, can past 20 years. Nevertheless, because of population growth University, Mannheim, Germany
(Prof J B Jonas MD);
affect the causes and prevalence of vision loss. The and the increase in the number of elderly adults, the LV Prasad Eye Institute,
information about vision loss, therefore, might not be blind population has remained stable and the vision- Hyderabad, India
relevant at the worldwide level. Assessment of causes of impaired population might have increased.12 (Prof J Keeffe PhD); College of
vision impairment, their trends, and the effects of Here we report a systematic analysis of the VLEG GBD Optometry, Nova Southeastern
University, Fort-Lauderdale-
interventions is most accurate when repeated surveys are dataset. We aimed to estimate the main causes of blindness Davie, FL, USA (J Leasher OD);
done within the same population, but such data are and vision impairment worldwide and by geographical African Vision Research
rarely collected. In their absence, estimates of patterns region, including analysis of trends over time. Institute, University of Kwazulu-
and trends in vision impairment derived from available Natal, South Africa and Brien
Holden Vision Institute, Sydney,
data might be useful to set policy priorities. Methods NSW, Australia (K Naidoo PhD);
Estimates of the leading causes of vision impairment Study design NHMRC Centre for Clinical Eye
worldwide have been generated by meta-analyses.5–8 The We estimated trends in causes of vision impairment, Research, Flinders University,
Adelaide, SA, Australia
first of these, done by Thylefors and colleagues in 1995,5 including analysis of uncertainties, by age, sex, and

www.thelancet.com/lancetgh Vol 1 December 2013 e339

 Articles

(Prof K Pesudovs PhD); geographical region (we used the 21 regions defined in previously;11 estimation of fractions for each cause, by
International Health and the GBD; appendix pp 1–2). We estimated what severity of vision impairment, sex, age, and region, as
Development, Geneva,
Switzerland
proportions of overall vision impairment were from six described in this report; and application of cause
(Prof S Resnikoff PhD); and common causes of blindness and vision impairment: fractions to the prevalence of all-cause presenting vision
Melbourne School of Public cataract, glaucoma, macular degeneration, diabetic impairment, which was estimated previously.12
Health, University of retinopathy, trachoma, and uncorrected refractive error
Melbourne, Melbourne, VIC,
Australia (Prof H R Taylor MD)
(table 1). We also estimated the proportions of blindness Data sources
and vision impairment related to other causes. We The methods for the data search have been published
Correspondence to:
Dr Rupert R A Bourne, Vision and calculated the causes of moderate and severe vision previously.11 Briefly, we searched for articles published
Eye Research Unit, East Road, impairment (MSVI), defined as visual acuity in the better from Jan 1, 1980, to Jan 31, 2012 (panel). Our initial
Anglia Ruskin University, eye lower than 6/18 but at least 3/60 at presentation, and search identified 14 908 relevant manuscripts, which
Cambridge CB1 1PT, UK
[email protected]
blindness, defined as visual acuity in the better eye lower were distilled by application of rigorous selection criteria
than 3/60 at presentation. We did our analysis in three and review by an expert panel to 243 high-quality,
See Online for appendix steps: data identification and access, as described population-based studies (appendix pp 3–18). Data from
epidemiological studies that reported prevalence
Description disaggregated by cause (128 studies) were used to
Cataract Age-related cataract leading to progressive visual impairment calculate the proportions of blindness and MSVI that
Glaucoma All types of glaucoma combined were due to cataract, glaucoma, macular degeneration,
Macular degeneration Degeneration of macula and posterior pole, including age-related macular diabetic retinopathy, trachoma, and undercorrection of
degeneration, myopic maculopathy, macular hole, and any other macular disorder refractive error or other causes. Many countries with
Diabetic retinopathy Diabetic retinopathy and sequelae endemic trachoma lack epidemiological data on
Trachoma Trachoma-related corneal scarring trachomatous blindness and, therefore, we also obtained
Uncorrected refractive Estimated as the difference between vision impairment at presentation and data from the authors of an unpublished working paper,
error best-corrected sight (includes aphakia) which uses trichiasis data extracted from the Global Atlas
Other All other causes, including unidentified causes of Trachoma13 to model the prevalence of trachoma-
related vision impairment in specific countries
Table 1: Definitions of causes of blindness and vision impairment used in this study
(Smith JL, Solomon A, Haddad D, Brooker S, personal
communication).
Panel: Research in context
Statistical analysis
Systematic review We used the DisMod-MR model from GBD to calculate
We did a systematic review of articles published from January, 1980, to January, 2012, in the fraction of vision impairment due to cataract,
the following sources: Medline, Embase, and the WHO library information system. Search glaucoma, macular degeneration, and diabetic retinopathy.
terms included concepts to describe “blindness”, “VI”, “population”, “eye”, “survey”, and a This model is a Bayesian multi-level regression tool that
list of disorders that affect the eyes. We identified additional unpublished data sources incorporates age.10,14 Briefly, DisMod-MR includes the
through personal communication with researchers identified in the literature search.11 following elements: covariates that predict variation in the
Literature reviews published by WHO and the WHO Prevention of Blindness and Deafness true proportion of vision impairment from each disease
programme have been used to make worldwide estimates of numbers of people blind or (eg, year); fixed effects that adjust for differences in
with vision impairment. The latest of studies included literature published in 2000–10, definitions (eg, whether causes were reported on the basis
and the analysis was limited to three age groups, with no breakdown by sex, provision of of vision at presentation vs best-corrected vision); a
a point estimate for 2010, or estimates for the six WHO epidemiological subregions hierarchical model structure that fits random intercepts in
within a more limited timeframe.8 individual countries derived from the data observed in a
Interpretation country, its region, and in other regions, on the basis of
We have previously reported prevalence of vision impairment and blindness worldwide and the availability and consistency of country-specific and
shown that the age-standardised prevalence of both has decreased between 1990 and region-specific data; a piecewise linear spline model of the
2010.12 This study added to those findings by investigating the contribution of various causes age pattern; and a fixed effect for data on men and boys.
to the burden of vision loss, and by analysing temporal trends in contribution. The We used a specific set of parameters for each cause of
proportions of vision impairment and blindness due to cataract and trachoma decreased over vision impairment (appendix p 19). We assessed all model
the study period of 20 years; those due to glaucoma, macular degeneration, diabetic fits visually.
retinopathy, and uncorrected refractive error increased. By contrast with the WHO data, we For causes of cataract we fitted two DisMod-MR
undertook a granular analysis, and present data in 5-year age groups and by sex, provide models, one with data on the proportion of blindness
time-series estimates for the period 1990–2010, and break them down geographically for caused by cataract and one for MSVI caused by cataract
21 regions. Thus, our estimates of prevalence are more detailed and show temporal change. (appendix p 19). We fitted each model with three
This analysis, therefore, provides useful information for the setting of priorities, development covariates: an indicator variable that described whether
of policies, and planning. Additionally, the data might provide a resource for advocacy efforts the data were based on visual acuity at presentation or
to help mobilise resources for eye-care services from governments, donors, and civil society. best-corrected visual acuity, a year covariate that allowed
for analysis of time trends, and a country covariate that

e340 www.thelancet.com/lancetgh Vol 1 December 2013

 Articles

reflected access to health systems.15 We have shown this regression model to predict the proportion of
previously that the latter variable, which was developed blindness and MSVI caused by trachoma in men and
for the GBD, can predict the prevalence of all-cause women in each endemic country, for 1990 and 2010.
vision impairment.12 We predicted the proportion of best- As described previously,12 the total prevalence of vision
corrected vision impairment that was caused by cataract. impairment and its uncertainty were estimated from
We fitted one DisMod-MR model for each of glaucoma, data on presenting visual acuity and best-corrected
macular degeneration, and diabetic retinopathy visual acuity. This model implicitly estimated the
(appendix p 19). In the models for glaucoma and macular difference between the prevalence of blindness (and of
degeneration we used three covariates: an indicator MSVI) on the basis of visual acuity and on best-corrected
variable that described whether the data were for visual acuity. We interpreted this difference as the
blindness or for MSVI, another that described whether prevalence of vision impairment caused by uncorrected
the data were based on visual acuity at presentation or refractive error.
best-corrected visual acuity, and a country covariate that The proportions of best-corrected vision attributable to
reflected access to health systems. We made two sets of causes other than cataract, macular degeneration,
predictions for glaucoma and macular degeneration, one glaucoma, or trachoma were calculated with data from
for best-corrected blindness and one for best-corrected surveys that included at least cataract and macular
MSVI. For diabetic retinopathy, we used three covariates: degeneration (only 3% of survey data did not report
an indicator variable that described whether the data glaucoma). We modelled these data with DisMod-MR
were based on visual acuity at presentation or best- (appendix p 19). The covariates were year and an indicator
corrected visual acuity, a year covariate to allow for time variable for MSVI. We made two sets of predictions, one
trends, and a country covariate that reflected access to for blindness and one for MSVI. We deducted the
health systems. We predicted the proportion best- estimated proportions of blindness and MSVI caused by
corrected visual acuity caused by diabetic retinopathy, diabetic retinopathy to find the proportion of best-
and used the same proportions for blindness and MSVI. corrected vision caused by other vision impairment. This
Estimates for the prevalence of trachoma were derived category comprises avoidable, unavoidable, and
from nationally representative surveys of vision unidentified causes.
impairment and from a Bayesian predictive model that DisMod-MR produced 1000 draws for each country,
used data on the prevalence of trichiasis, a clinical stage sex, age, and year (1990 and 2010). We used the mean of
of trachoma that is a direct cause of visual impairment, the draws as the central estimate. For computational
as described above. In 16 countries we obtained data on efficiency, we selected every second draw (total 500) to
the proportion of vision impairment caused by trachoma propagate uncertainty. We also extracted the central
from national surveys, in 25 countries we based estimate and 500 draws for uncorrected refractive error.12
estimates on trichiasis prevalence, and in two countries For the central estimate and each draw, we normalised
we used both sources of data. On the basis of national the proportions attributable to all causes of best-corrected
survey data, we made estimates for five countries that vision loss (cataracts, glaucoma, macular degeneration,
WHO does not currently classify as having endemic diabetic retinopathy, trachoma, and other) to sum to the
disease: Dominican Republic, Ecuador, Paraguay, Saudi remaining vision impairment not due to refractive error.
Arabia, and Thailand.16 For another 20 countries that are We applied the results to previously estimated prevalence
deemed by WHO to be endemic for trachoma16 we could of blindness and MSVI by country, age, sex, and draw.12
obtain no data and conservatively assigned each as We calculated uncertainty intervals (UIs) as the
having no trachoma-related vision impairment. These 2·5th–97·5th percentiles of the distribution of draws.
countries were predominantly small countries that For presentation, we age-standardised prevalence with
would not affect regional or worldwide estimates (eg, the the WHO reference population.17 We also calculated
Solomon Islands) or those that had no data because the numbers of people with blindness and MSVI by cause
prevalence of trachomatous blindness is low (eg, in each region to reflect that region’s population size
Pakistan). For all countries classified as not having and age distribution.
endemic trachoma, we estimated no trachoma-related
vision impairment. Role of the funding sources
With these data, we fitted the following regression The sponsors had no involvement in study design, data
(equation, appendix p 19): collection, data analysis, data interpretation, or writing of
the report. The corresponding author had full access to
logit(PT) = β0 + β1sex + β2year + β3vision level + β4country all the data in the study and had final responsibility for
the decision to submit the paper for publication.
Vision level was an indicator variable used to indicate
whether data related to blindness or MSVI. Country was Results
a fixed effect, sex was an indicator variable, and year was At least two studies were identified for 18 of 21 GBD
a linear covariate. We used the fitted coefficients from study regions. No studies with cause-specific data were

www.thelancet.com/lancetgh Vol 1 December 2013 e341

 Articles

identified for central Africa or eastern Europe and one In 1990 and in 2010, the leading causes of blindness
study was identified for central Europe. No studies were worldwide, based on presenting visual acuity
identified for 126 (66%) of 191 countries. measurements, were cataract, uncorrected refractive

Cataract Uncorrected Macular Glaucoma Diabetic Trachoma Other causes/

refractive error degeneration retinopathy unidentified
1990
Asia Pacific, high income 18·8% (13·3–25·6) 14·0% (8·4–18·2) 14·9% (10·3–21·3) 9·0% (6·0–12·8) 4·8% (3·3–7·2) 0 38·5% (30·4–46·4)
Asia, central 29·0% (24·3–33·6) 13·7% (8·2–17·6) 11·6% (8·8–15·3) 9·5% (7·3–12·6) 3·4% (2·7–4·6) 0 33·0% (28·7–37·5)
Asia, east 37·2% (29·7–45·4) 13·5% (8·0–17·5) 5·0% (3·2–7·9) 3·9% (2·6–5·8) 1·0% (0·65–1·6) 5·4% (4·4–6·6) 34·1% (26·5–42·6)
Asia, south 47·7% (39·5–59·4) 35·4% (20·3–45·9) 1·4% (1·0–1·9) 2·4% (1·7–3·3) 1·9% (1·3–2·9) 0·25% (0·20–0·35) 10·9% (8·6–13·9)
Asia, southeast 47·2% (42·0–51·7) 13·0% (7·8–17·1) 3·7% (2·9–5·0) 3·3% (2·6–4·4) 1·1% (0·88–1·4) 0·33% (0·25–0·45) 31·4% (27·2–36·1)
Australasia 19·7% (15·2–26·0) 14·0% (8·4–18·0) 16·8% (12·9–22·2) 9·6% (7·4–13·1) 4·5% (3·4–6·2) 0 35·5% (29·2–41·6)
Caribbean 32·9% (28·3–38·0) 13·3% (8·0–17·4) 4·5% (3·4–6·0) 9·1% (7·3–11·8) 2·1% (1·7–2·8) 0·03% (0·02–0·04) 38·0% (33·4–42·7)
Europe, central 26·9% (23·1–31·5) 13·8% (8·2–17·9) 12·2% (9·3–15·6) 10·2% (7·9–13·3) 3·5% (2·8–4·5) 0 33·4% (28·6–38·3)
Europe, eastern 25·3% (18·7–32·5) 13·8% (8·3–17·7) 13·1% (9·1–19·2) 10·8% (7·7–15·4) 3·5% (2·4–5·4) 0 33·7% (25·5–42·0)
Europe, western 19·2% (16·2–22·7) 13·9% (8·3–18·0) 16·1% (13·4–19·6) 9·0% (7·4–11·3) 4·4% (3·8–5·6) 0 37·4% (32·4–43·0)
Latin America, Andean 37·5% (30·5–44·2) 13·3% (8·0–17·4) 3·7% (2·6–5·5) 6·8% (5·0–9·6) 2·1% (1·5–3·1) 0 36·6% (30·4–43·0)
Latin America, central 32·9% (27·6–38·6) 13·2% (8·0–17·2) 4·6% (3·5–6·2) 8·6% (6·7–11·7) 2·2% (1·7–2·9) 0 38·6% (32·6–43·8)
Latin America, southern 24·3% (18·5–31·2) 13·6% (8·1–17·6) 14·6% (10·6–19·6) 9·3% (6·6–12·6) 5·4% (3·9–8·0) 0 32·9% (25·0–40·1)
Latin America, tropical 32·7% (23·9–42·5) 13·4% (8·0–17·3) 5·0% (3·1–8·0) 9·2% (5·9–14·0) 2·5% (1·6–4·1) 0 37·2% (29·0–46·1)
North Africa/Middle East 29·2% (25·5–33·4) 12·7% (7·6–16·6) 6·4% (5·2–8·0) 5·6% (4·4–7·6) 2·7% (2·3–3·5) 5·1% (3·4–6·2) 38·3% (34·2–42·8)
North America, high income 17·4% (12·5–22·8) 14·0% (8·4–18·1) 16·4% (12·2–21·4) 9·2% (6·7–11·9) 4·1% (3·0–5·8) 0 39·0% (32·0–46·9)
Oceania 43·3% (35·2–49·9) 13·4% (8·0–17·5) 3·3% (2·3–4·7) 2·8% (2·0–4·1) 1·2% (0·85–1·8) 0 36·0% (29·7–43·7)
Sub-Saharan Africa, central 41·0% (33·3–47·5) 13·3% (7·8–17·2) 4·8% (3·5–7·1) 3·3% (2·4–4·6) 2·5% (1·8–3·6) 0·94% (0·52–1·6) 34·2% (28·4–41·2)
Sub-Saharan Africa, east 35·4% (31·7–39·8) 12·9% (7·7–17·1) 4·1% (3·4–5·1) 2·9% (2·4–3·6) 2·0% (1·6–2·5) 13·5% (11·7–15·1) 29·2% (26·4–32·5)
Sub-Saharan Africa, southern 34·0% (29·0–39·8) 13·2% (7·8–17·3) 6·9% (5·5–8·9) 5·4% (4·2–7·3) 2·9% (2·0–4·1) 1·6% (1·1–2·6) 36·1% (30·1–42·4)
Sub-Saharan Africa, west 37·1% (32·1–41·9) 12·9% (7·6–17·0) 4·1% (3·4–5·4) 2·9% (2·4–3·8) 2·4% (1·9–3·1) 7·3% (6·6–8·6) 33·4% (28·9–38·4)
Worldwide 38·6% (35·2–42·0) 19·9% (14·9–24·9) 4·9% (4·4–5·8) 4·4% (4·0–5·1) 2·1% (1·9–2·5) 2·8% (2·3–3·1) 27·4% (24·9–30·0)
2010
Asia Pacific, high income 13·1% (8·3–20·8) 14·1% (8·4–18·2) 19·5% (12·3–28·8) 11·7% (7·1–18·8) 4·3% (2·6–7·1) 0 37·3% (25·9–46·7)
Asia, central 24·2% (18·6–29·7) 13·9% (8·3–18·0) 13·3% (9·4–18·1) 12·0% (8·7–17·2) 4·0% (2·9–6·0) 0 32·6% (26·4–38·9)
Asia, east 28·2% (19·3–37·3) 13·8% (8·2–17·8) 6·9% (4·5–11·0) 5·4% (3·5–8·5) 1·1% (0·61–2·0) 2·0% (1·6–2·5) 42·7% (33·4–51·3)
Asia, south 41·7% (33·0–51·6) 36·0% (20·7–46·6) 2·6% (1·7–4·2) 4·7% (3·3–7·5) 2·8% (1·7–4·8) 0·15% (0·11–0·24) 12·1% (9·1–17·3)
Asia, southeast 42·0% (34·8–47·9) 13·4% (8·0–17·4) 5·9% (4·7–8·3) 5·6% (4·3–8·2) 1·4% (1·1–2·1) 0·15% (0·11–0·23) 31·5% (25·8–37·3)
Australasia 14·5% (8·5–22·4) 14·1% (8·4–18·1) 17·7% (11·1–26·3) 11·3% (6·8–18·8) 4·3% (2·5–7·7) 0 38·2% (25·9–48·1)
Caribbean 30·2% (23·8–37·1) 13·5% (8·1–17·8) 6·1% (4·3–8·9) 11·2% (8·0–15·1) 2·3% (1·7–3·4) 0·01% (0·01–0·02) 36·8% (30·7–42·7)
Europe, central 21·6% (17·0–26·7) 14·0% (8·3–18·0) 15·4% (10·9–20·8) 12·5% (9·1–17·0) 3·7% (2·8–5·5) 0 32·9% (26·7–38·3)
Europe, eastern 20·6% (13·1–30·5) 14·0% (8·4–18·0) 16·6% (10·1–25·1) 13·5% (8·6–20·6) 4·0% (2·5–6·9) 0 31·3% (21·0–40·9)
Europe, western 13·8% (11·2–17·9) 14·0% (8·4–18·1) 16·1% (12·5–20·1) 10·6% (8·2–14·0) 4·2% (3·4–5·9) 0 41·4% (35·4–46·5)
Latin America, Andean 31·0% (23·3–38·7) 13·6% (8·2–17·7) 6·2% (4·1–9·2) 11·7% (7·9–17·1) 2·5% (1·6–4·2) 0 35·1% (26·7–43·0)
Latin America, central 26·4% (20·9–32·3) 13·5% (8·1–17·5) 6·8% (4·9–9·9) 13·0% (9·6–18·2) 2·5% (1·9–3·7) 0 37·8% (30·4–44·6)
Latin America, southern 18·0% (12·0–25·8) 13·7% (8·2–17·7) 19·5% (13·2–26·8) 12·6% (7·9–19·3) 5·5% (3·6–9·1) 0 30·8% (21·8–39·3)
Latin America, tropical 23·9% (16·2–32·5) 13·6% (8·1–17·5) 9·1% (5·7–13·3) 15·5% (9·6–21·9) 2·9% (1·9–4·6) 0 35·1% (26·3–43·7)
North Africa/Middle East 23·4% (18·7–28·6) 13·1% (7·8–17·1) 10·3% (7·8–13·6) 9·6% (7·5–13·2) 3·5% (2·8–5·0) 2·6% (1·6–3·3) 37·6% (31·6–43·3)
North America, high income 12·7% (8·3–18·7) 14·1% (8·4–18·2) 16·4% (10·8–23·4) 10·7% (7·0–15·7) 3·9% (2·7–6·4) 0 42·2% (32·6–50·1)
Oceania 40·6% (31·5–48·6) 13·6% (8·1–17·7) 4·6% (3·1–7·6) 4·2% (2·5–7·2) 1·4% (0·91–2·4) 0 35·6% (27·4–45·0)
Sub-Saharan Africa, central 34·8% (25·3–42·5) 13·6% (8·0–17·5) 6·9% (4·7–11·0) 5·2% (3·4–8·8) 3·0% (2·0–5·2) 0·44% (0·24–0·82) 36·0% (28·4–44·0)
Sub-Saharan Africa, east 36·7% (31·9–41·5) 13·1% (7·8–17·2) 5·8% (4·6–7·7) 4·0% (3·1–5·4) 2·4% (1·9–3·4) 8·1% (6·8–9·5) 29·9% (25·6–34·5)
Sub-Saharan Africa, southern 31·2% (24·6–39·0) 13·5% (8·0–17·7) 9·7% (6·7–14·1) 7·3% (5·2–10·4) 3·4% (2·1–5·8) 0·69% (0·45–1·1) 34·0% (27·1–42·2)
Sub-Saharan Africa, west 33·8% (28·1–39·3) 13·2% (7·8–17·3) 6·2% (4·8–8·4) 4·4% (3·4–5·9) 3·1% (2·4–4·9) 3·6% (3·2–4·6) 35·6% (30·1–41·6)
Worldwide 33·4% (29·6–36·4) 20·9% (15·2–25·9) 6·6% (6·0–7·9) 6·6% (5·9–7·9) 2·6% (2·2–3·4) 1·4% (1·2–1·7) 28·6% (26·1–31·5)

Data are proportions (95% uncertainty intervals).

Table 2: Causes of blindness in 21 regions and worldwide in 1990 and 2010

e342 www.thelancet.com/lancetgh Vol 1 December 2013

 Articles

5 Worldwide Cataracts
Macular degeneration
4

Prevalence (%)
Glaucoma
3 Trachoma
Diabetic retinopathy
2
Refractive error
1 Other
0

5 Asia Pacific, high income 5 North America, high income 5 Australasia

4 4 4
Prevalence (%)

3 3 3
2 2 2
1 1 1
0 0 0

5 Western Europe 5 Central Europe 5 Eastern Europe

4 4 4
Prevalence (%)

3 3 3
2 2 2
1 1 1
0 0 0

5 Central Asia 5 Southern Latin America 5 East Asia

4 4 4
Prevalence (%)

3 3 3
2 2 2
1 1 1
0 0 0

5 Tropical Latin America 5 Caribbean 5 Andean Latin America

4 4 4
Prevalence (%)

3 3 3
2 2 2
1 1 1
0 0 0

5 Central Latin America 5 Oceania 5 Central Africa

4 4 4
Prevalence (%)

3 3 3
2 2 2
1 1 1
0 0 0

5 Southern Africa 5 Southeast Asia 20 South Asia

4 4
Prevalence (%)

15
3 3
10
2 2
1 1 5

0 0 0

20 East Africa 20 West Africa 20 North Africa/Middle East

Prevalence (%)

15 15 15

10 10 10

5 5 5

0 0 0
1990 1995 2000 2005 2010 1990 1995 2000 2005 2010 1990 1995 2000 2005 2010
Years Years Years

Figure 1: Prevalence of blindness in adults aged 50 years and older, by cause, in 21 regions and worldwide, from 1990 to 2010

www.thelancet.com/lancetgh Vol 1 December 2013 e343

 Articles

error, and macular degeneration (table 2, figure 1). men. The disparities between sexes were less for all other
Avoidable vision loss due to preventable or treatable causes of vision impairment.
causes can be defined as vision loss due to cataract, The age-standardised prevalence of trachoma, cataract,
uncorrected refractive error, trachoma, glaucoma, and and uncorrected refractive error worldwide between 1990
diabetic retinopathy. Avoidable vision loss due to and 2010 showed the greatest declines (appendix
preventable or treatable causes affected 68% (95% UI pp 20–23). For glaucoma, macular degeneration, and
65–70) of 31·8 million blind people in 1990, but by 2010, diabetic retinopathy, prevalence had declined less (for
the proportion had decreased significantly to 65% (61–68) blindness) or increased slightly (for MSVI).
of 32·4 million blind. Additionally, the rate of preventable The number of people affected by blindness due
or treatable MSVI decreased from 1990 (80%, 78–83) to to cataract decreased between 1990 and 2010 from
2010 (76%, 73–79). 12·3 million (95% UI 10·7 million to 14·2 million) to
Breakdown of the worldwide averages showed large 10·8 million (9·3 million to 12·3 million), and for MSVI
differences in the causes of blindness between regions fell from 44·0 million (35·6 million to 52·4 million) to
(figure 1). In 2010, the proportion of blindness caused by 35·2 million (29·6 million to 43·5 million; appendix
cataracts ranged from less than 15%, with the lowest pp 20–23). In 2010, the age-standardised prevalence of
values seen in high-income regions, to more than 40% in blindness and MSVI caused by cataract in people aged
south and southeast Asia and Oceania. The proportion of 50 years or older was 0·7% (95% UI 0·6–0·8) and 2·2%
blindness caused by macular degeneration was higher in (1·9–2·7), respectively. These values represent declines
regions with older populations, such as high-income from 1990 (1·3%, 1·1–1·5 for blindness and 4·4%, 3·6–5·2
regions and southern Latin America, and central and for MSVI; appendix pp 20–23). The decline in blindness or
eastern Europe, where more than 15% of blindness was MVSI due to cataract was greatest in east Asia, tropical
caused by macular degeneration, whereas the proportion Latin America, and western Europe, in all of which
was much lower in regions such as south Asia (2·6%, prevalence fell by more than half. The region with the least
95% UI 1·7–4·2; table 2). The proportion of blindness decline was east sub-Saharan Africa (appendix pp 24–27).
caused by glaucoma varied notably, with the lowest The number of people with blindness or MSVI caused
values being seen in south Asia (4·7%, 3·3–7·5), east and by trachoma decreased from 0·87 million (95% CI
west sub-Saharan Africa (4·0%, 3·1–5·4 and 4·4%, 0·70 million to 1·0 million) and 2·2 million (1·5 million
3·4–5·9, respectively), and Oceania (4·2%, 2·5–7·2), and to 2·8 million) in 1990 to 0·45 million (0·38 million to
the highest value being seen in tropical Latin America 0·54 million) and 1·4 million (1·1 million to 1·8 million)
(15·5%, 9·6–21·9). We estimated that there was no in 2010, respectively. The age-standardised prevalence of
trachoma-related blindness in 13 of 21 world regions, but trachoma as a cause of blindness and MSVI combined
that 3·6% (3·2–4·6) of blindness in west sub-Saharan was 0·29% (0·22–0·34) in 1990 and 0·11% (0·09–0·13)
Africa and 8·1% (6·8–9·5) in east sub-Saharan Africa in 2010. The proportion of global blindness caused by
was caused by trachoma-related corneal scars in 2010. trachoma decreased from 2·8% in 1990 (2·3–3·1) to
Worldwide, the leading causes of MSVI were 1·4% (1·2–1·7) in 2010; for MSVI the values are 1·3%
uncorrected refractive error, cataract, and macular (1·0–1·5) and 0·7% (0·6–0·9), respectively. The greatest
degeneration (table 3, figure 2). In 2010, uncorrected decreases were seen in east and west sub-Saharan Africa
refractive error caused a larger proportion of MSVI in (appendix pp 24–27).
south Asia (65·4%, 95% UI 62·0–72·0) than in other The number of people worldwide affected by blindness
regions (range 43·2–48·1%). As with blindness, the or MSVI caused by uncorrected refractive error increased
proportion of MSVI caused by cataract was smallest in from 6·3 million (4·4 million to 8·1 million) and
the highest-income regions (range 13·0–13·8%) and 88·0 million (69·9 million to 103·3 million) in 1990 to
largest in south Asia (21·4%, 95% UI 16·1–24·2) and 6·8 million (4·7 million to 8·8 million) and 101·2 million
southeast Asia (22·7%, 17·9–27·4). The proportion of (87·88 million to 125·5 million) in 2010, respectively. The
MSVI caused by macular degeneration was small in age-standardised prevalence of uncorrected refractive
comparison (range 1·0–8·0%). Glaucoma, diabetic error as a cause for adult blindness and MSVI combined,
retinopathy, and trachoma caused less than 5·5% of however, was 7·5% (6·1–8·5%) in 1990 and 5·7%
MSVI in all regions. (5·0–6·9%) in 2010. Between 1990 and 2010, the
In all regions and worldwide, higher proportions of percentage reductions in age-standardised prevalence of
blindness and MSVI were caused by cataract and macular uncorrected refractive error as a cause for adult blindness
degeneration in women than in men. Worldwide, and MSVI combined were greatest in tropical Latin
cataract caused 35·5% (95% UI 31·0–39·1) of blindness America (36%), central Asia (36%), and high-income
in women, compared with 30·1% (25·2–33·7) in men, Asia Pacific (35%), and smallest in eastern sub-Saharan
and for MSVI the values were 20·2% (17·2–23·0) and Africa (17%), Oceania (20%), and western sub-Saharan
15·9% (12·8–18·6), respectively. Likewise, macular Africa (21%).
degeneration caused 7·3% (6·4–8·9) of blindness in The worldwide age-standardised prevalence for
women worldwide, compared with 5·5% (4·8–6·8) in blindness and for MSVI declined substantially from 1990

e344 www.thelancet.com/lancetgh Vol 1 December 2013

 Articles

to 2010. Of this overall decline in vision impairment, reductions in the prevalence of blindness and MSVI,
around half was a result of decline in vision impairment respectively, resulted from declines in uncorrected
caused by cataracts. A further 20% and 45% of the refractive error. Despite the large decline in the

Cataract Uncorrected Macular Glaucoma Diabetic Trachoma Other causes/

refractive error degeneration retinopathy unidentified
1990
Asia Pacific, high income 22·8% (18·1–29·4) 47·9% (38·4–54·1) 3·5% (2·4–5·1) 2·3% (1·5–3·5) 2·8% (1·9–4·1) 0 20·8% (15·8–27·3)
Asia, central 26·6% (22·1–31·2) 45·1% (36·8–51·6) 3·3% (2·4–4·6) 2·3% (1·8–3·3) 2·0% (1·6–3·0) 0 20·6% (17·3–25·1)
Asia, east 24·1% (18·9–29·4) 44·8% (36·4–51·7) 3·1% (1·9–5·2) 0·92% (0·57–1·5) 0·68% (0·43–1·1) 2·9% (2·3–3·8) 23·6% (17·8–29·3)
Asia, south 26·7% (21·4–29·5) 63·9% (60·0–70·8) 0·45% (0·32–0·61) 0·66% (0·47–0·90) 1·2% (0·76–1·8) 0·13% (0·10–0·17) 7·0% (5·4–8·6)
Asia, southeast 30·1% (25·4–35·2) 42·8% (34·2–49·7) 0·93% (0·72–1·3) 0·83% (0·65–1·1) 0·76% (0·59–1·0) 0·20% (0·15–0·28) 24·4% (20·4–29·4)
Australasia 21·9% (17·2–27·6) 46·5% (37·4–53·1) 6·6% (4·6–9·5) 2·4% (1·7–3·4) 2·6% (1·9–3·6) 0 20·1% (15·8–25·7)
Caribbean 22·8% (18·4–27·4) 43·4% (35·6–50·4) 0·74% (0·57–1·0) 3·0% (2·2–4·1) 1·5% (1·2–2·1) 0·02% (0·01–0·03) 28·6% (23·8–34·0)
Europe, central 26·6% (21·9–31·7) 45·2% (36·1–52·1) 4·7% (3·4–6·5) 2·5% (1·9–3·3) 2·0% (1·5–2·7) 0 19·1% (15·8–23·4)
Europe, eastern 26·4% (20·4–32·1) 44·8% (36·6–51·3) 4·3% (2·7–6·9) 2·7% (1·8–4·2) 2·0% (1·4–3·2) 0 19·8% (14·8–26·0)
Europe, western 24·1% (19·9–28·8) 46·6% (37·9–53·0) 4·1% (3·3–5·2) 2·3% (1·8–3·0) 2·5% (2·0–3·3) 0 20·5% (17·2–24·5)
Latin America, Andean 22·9% (17·8–28·6) 43·4% (35·0–50·8) 1·4% (0·99–2·2) 2·1% (1·4–3·0) 1·6% (1·1–2·4) 0 28·7% (23·4–34·8)
Latin America, central 21·6% (17·3–26·0) 43·9% (35·6–50·8) 1·7% (1·2–2·5) 2·5% (1·9–3·7) 1·6% (1·2–2·2) 0 28·7% (24·0–34·4)
Latin America, southern 26·5% (21·1–32·6) 44·4% (35·6–51·5) 4·2% (2·9–6·2) 2·4% (1·7–3·7) 3·1% (2·3–5·0) 0 19·3% (14·6–24·7)
Latin America, tropical 22·0% (16·0–28·3) 44·2% (36·0–50·9) 2·8% (1·7–4·6) 2·7% (1·6–4·5) 1·7% (1·1–2·8) 0 26·6% (20·5–32·8)
North Africa/Middle East 25·1% (20·8–29·4) 41·4% (33·1–49·2) 1·8% (1·4–2·6) 1·4% (1·1–1·9) 1·6% (1·3–2·1) 3·7% (2·0–5·0) 25·0% (21·3–29·7)
North America, high income 21·1% (16·6–26·1) 47·8% (38·7–54·3) 4·5% (3·2–6·1) 2·4% (1·7–3·4) 2·4% (1·8–3·5) 0 21·9% (17·3–27·8)
Oceania 25·2% (20·1–31·5) 43·5% (35·4–50·6) 1·4% (0·99–2·1) 0·73% (0·51–1·1) 0·82% (0·56–1·3) 0 28·3% (22·7–33·9)
Sub-Saharan Africa, central 27·1% (22·2–32·4) 44·5% (35·5–51·3) 1·9% (1·3–2·9) 1·0% (0·71–1·5) 1·7% (1·2–2·5) 0·46% (0·20–0·85) 23·4% (19·6–28·6)
Sub-Saharan Africa, east 22·7% (19·3–26·6) 44·0% (35·4–50·6) 2·5% (1·9–3·2) 0·95% (0·76–1·2) 1·3% (1·0–1·7) 8·0% (6·5–9·6) 20·6% (17·3–24·6)
Sub-Saharan Africa, southern 24·2% (19·3–29·6) 45·9% (36·7–52·6) 2·8% (2·0–4·2) 1·5% (1·1–2·3) 1·7% (1·3–2·6) 0·85% (0·40–1·6) 22·9% (18·4–28·6)
Sub-Saharan Africa, west 24·3% (20·0–28·2) 43·8% (35·0–50·4) 1·4% (1·1–1·9) 0·93% (0·71–1·3) 1·6% (1·3–2·3) 4·3% (3·7–5·3) 23·7% (19·7–29·1)
Worldwide 25·6% (22·7–28·4) 51·1% (45·6–56·0) 1·9% (1·6–2·4) 1·2% (1·1–1·5) 1·3% (1·2–1·6) 1·3% (0·97–1·5) 17·6% (15·4–20·3)
2010
Asia Pacific, high income 15·2% (8·2–22·1) 48·1% (38·6–54·4) 6·0% (3·8–9·8) 3·7% (2·3–6·9) 3·1% (2·0–5·4) 0 23·9% (17·3–30·6)
Asia, central 18·7% (13·9–23·5) 46·5% (37·2–52·7) 5·0% (3·5–7·5) 3·6% (2·6–5·4) 2·8% (2·1–4·6) 0 23·4% (19·4–28·5)
Asia, east 13·4% (8·0–19·7) 46·1% (37·2–52·8) 5·2% (3·3–8·0) 1·6% (0·94–2·5) 0·84% (0·49–1·6) 1·2% (0·99–1·7) 31·6% (25·2–39·0)
Asia, south 21·4% (16·1–24·2) 65·4% (62·0–72·0) 1·0% (0·7–1·6) 1·6% (1·0–2·6) 2·1% (1·2–3·7) 0·08% (0·06–0·14) 8·6% (6·0–11·6)
Asia, southeast 22·7% (17·9–27·4) 44·2% (35·2–50·7) 1·8% (1·3–2·7) 1·8% (1·3–3·0) 1·2% (0·91–1·9) 0·10% (0·07–0·17) 28·1% (22·8–34·2)
Australasia 13·7% (8·4–20·8) 47·1% (37·7–53·5) 8·0% (5·1–12·6) 3·2% (1·9–5·9) 2·9% (1·8–5·9) 0 25·1% (17·4–32·4)
Caribbean 15·9% (11·4–21·3) 44·6% (36·3–51·2) 1·2% (0·90–1·9) 4·3% (3·1–6·4) 2·0% (1·5–3·2) 0·01% (0·00–0·02) 32·0% (25·7–38·4)
Europe, central 18·1% (13·2–23·4) 46·2% (36·7–52·7) 7·4% (5·2–11·0) 3·9% (2·8–6·0) 2·5% (1·8–4·0) 0 22·0% (17·5–27·0)
Europe, eastern 18·4% (11·8–26·2) 46·1% (37·1–52·6) 6·8% (4·0–10·9) 4·5% (2·6–7·7) 2·8% (1·8–5·2) 0 21·3% (14·4–29·1)
Europe, western 13·8% (10·3–18·3) 47·3% (38·5–53·7) 5·4% (4·1–7·5) 3·4% (2·5–4·9) 3·0% (2·4–4·5) 0 27·1% (22·5–32·9)
Latin America, Andean 14·8% (10·2–20·1) 44·6% (36·0–51·6) 3·1% (2·0–5·1) 4·5% (2·9–7·5) 2·2% (1·5–3·8) 0 30·8% (24·0–38·2)
Latin America, central 13·9% (9·9–18·8) 45·2% (36·2–51·6) 3·2% (2·2–5·0) 4·6% (3·2–7·1) 2·1% (1·6–3·5) 0 30·9% (24·7–37·3)
Latin America, southern 17·3% (10·7–23·2) 45·4% (36·5–52·0) 7·2% (4·5–11·5) 4·0% (2·5–6·3) 4·0% (2·6–6·8) 0 22·1% (15·9–28·7)
Latin America, tropical 13·9% (8·5–20·7) 45·4% (36·7–51·9) 6·0% (3·7–9·2) 5·2% (3·2–8·4) 2·2% (1·4–3·6) 0 27·3% (20·3–34·0)
North Africa/Middle East 18·0% (13·3–22·6) 43·2% (34·5–50·1) 4·1% (3·0–6·2) 3·0% (2·1–4·7) 2·4% (1·8–3·9) 2·1% (1·1–3·1) 27·1% (21·8–32·5)
North America, high income 13·0% (7·8–19·5) 48·1% (38·9–54·4) 5·5% (3·5–8·0) 3·4% (2·3–5·4) 2·8% (1·9–4·7) 0 27·3% (21·4–33·8)
Oceania 18·2% (12·1–25·4) 44·5% (35·8–51·3) 2·7% (1·7–4·8) 1·4% (0·85–2·5) 1·2% (0·77–2·3) 0 32·0% (24·2–40·1)
Sub-Saharan Africa, central 18·8% (12·8–24·3) 45·9% (36·8–52·4) 3·6% (2·1–6·2) 1·9% (1·2–3·3) 2·3% (1·6–4·0) 0·25% (0·10–0·53) 27·3% (21·4–35·4)
Sub-Saharan Africa, east 19·6% (15·8–23·6) 44·8% (36·0–51·0) 4·0% (3·0–5·5) 1·5% (1·1–2·2) 1·8% (1·4–2·6) 5·3% (4·2–6·9) 23·1% (19·0–28·0)
Sub-Saharan Africa, southern 17·8% (12·3–23·9) 46·7% (37·4–53·2) 4·8% (2·8–7·7) 2·6% (1·8–4·0) 2·5% (1·6–4·6) 0·47% (0·24–0·96) 25·2% (19·7–32·4)
Sub-Saharan Africa, west 15·6% (11·4–20·5) 44·8% (35·8–51·2) 2·9% (2·0–4·3) 1·8% (1·3–2·7) 2·7% (1·9–4·5) 2·5% (2·1–3·5) 29·6% (23·9–35·3)
Worldwide 18·4% (15·8–20·9) 52·9% (47·2–57·3) 3·1% (2·7–4·0) 2·2% (2·0–2·8) 1·9% (1·6–2·7) 0·71% (0·56–0·91) 20·8% (18·4–23·8)

Data are proportions (95% uncertainty intervals).

Table 3: Causes of moderate and severe vision impairment in 21 regions and worldwide in 1990 and 2010

www.thelancet.com/lancetgh Vol 1 December 2013 e345

 Articles

5 Worldwide Cataracts
Macular degeneration
4

Prevalence (%)
Glaucoma
3 Trachoma
Diabetic retinopathy
2
Refractive error
1 Other
0

5 Asia Pacific, high income 5 North America, high income 20 Western Europe
4 4
Prevalence (%)

15
3 3
10
2 2
1 1 5

0 0 0

20 Australasia 20 Central Europe 20 Southern Africa

Prevalence (%)

15 15 15

10 10 10

5 5 5

0 0 0

20 Central Asia 20 Eastern Europe 20 East Asia

Prevalence (%)

15 15 15

10 10 10

5 5 5

0 0 0

20 Southern Latin America 20 Tropical Latin America 20 Caribbean

Prevalence (%)

15 15 15

10 10 10

5 5 5

0 0 0

20 Central Africa 20 Central Latin America 20 East Africa

Prevalence (%)

15 15 15

10 10 10

5 5 5

0 0 0

20 Andean Latin America 20 West Africa 20 Southeast Asia

Prevalence (%)

15 15 15

10 10 10

5 5 5

0 0 0

20 Oceania 50 North Africa and Middle East 50 South Asia

40 40
Prevalence (%)

15
30 30
10
20 20
5 10 10
0 0 0
1990 1995 2000 2005 2010 1990 1995 2000 2005 2010 1990 1995 2000 2005 2010
Years Years Years

Figure 2: Prevalence of moderate and severe vision impairment in adults aged 50 years and older, by cause, in 21 regions and worldwide, from 1990 to 2010

e346 www.thelancet.com/lancetgh Vol 1 December 2013

 Articles

prevalence of blindness and vision impairment caused retinopathy caused only a few cases. Studies in Latin
by trachoma-related corneal scarring, it contributed only America, central India, and east Asia also confirm the
5% to overall decline worldwide in blindness and 3% to large proportion of presenting vision impairment due to
overall decline worldwide in MSVI. Nevertheless, the uncorrected refractive errors (eg, 30% in Ecuador and
reduction in corneal scarring from trachoma did make 72% in Brazil).19–22
important contributions to decline in vision impairment This study is an extension of a series of meta-analysis
in some regions, most notably in the east sub-Saharan investigations that started with a study by Thylefors and
Africa region, where it contributed 30% to decline in colleagues,5 which formed the basis for the 1999 launch
blindness and 21% to the decline in MSVI. of the Global Initiative for the Elimination of Avoidable
Blindness, also known as VISION 2020: the Right to
Discussion Sight.5 Since the publication of these worldwide data on
We found that the major causes of blindness in 2010 blindness in 1995, population-based studies on the
were, in order, cataract, uncorrected refractive error, and prevalence of blindness and vision impairment have
macular degeneration and for MSVI were uncorrected been done in all WHO regions. Nevertheless, there
refractive error, cataract, and macular degeneration. remains a dearth of such information from regions such
Thus, despite the numbers of people affected by blindness as central Africa, central and eastern Europe, and the
and MSVI due to cataract and trachoma decreasing and Caribbean.11 Our study expands the data derived from a
the number of people with uncorrected refractive error literature review published by WHO, which used the
and age-related diseases, such as glaucoma, macular WHO program Prevention of Blindness and Deafness
degeneration, and diabetic retinopathy, increasing, the Programme for estimates.8 In that study, data from 2000
leading causes were the same as in 1990. Causes of to 2010 were assessed, and the analysis was limited to
blindness differed substantially by region, with the three age groups, with no breakdown by sex, and
prevalence of cataract being lowest and that of macular provided a point estimate for 2010 and estimates for the
degeneration being greatest in highest-income regions. six WHO epidemiological world subregions.8 By contrast,
Worldwide and in all regions, the proportion of blindness we achieved a greater degree of granularity in our
or MSVI caused by cataract and macular degeneration analysis by presenting data in 5-year age groups and by
was higher in women than in men. sex, we calculated time-series estimates for the period
Avoidable vision loss due to preventable or treatable 1990–2010, and make estimates for 21 regions. Thus, our
causes can be defined as any vision loss due to cataract, estimates of prevalence of vision impairment have
uncorrected refractive error, trachoma, glaucoma, and increased detail and show temporal changes.
diabetic retinopathy. With this definition, of the The design of our study has potential limitations. First,
31·8 million people blind in 1990, 68% (95% UI 65–70) as in our previous study on the global prevalence of vision
had preventable or treatable causes. By 2010, the loss,12 a major limitation was that data were unavailable or
proportion had decreased to 65% (61–68) of 32·4 million those that were available were at a subnational level for
blind, which was a significant change. Some of the many country-years (appendix pp 3–18). Only a few
blindness assigned to other causes, which comprised national studies reported vision loss for all ages and all
29% (26–31) of blindness in 2010, might also be causes. Second, some data sources did not report
preventable or treatable. For example, onchocerciasis is prevalence by age (appendix pp 3–18). To use these data,
among the infectious preventable causes of blindness. we imputed age-specific proportions for causes, based on
Additionally, 76% (73–79) of MSVI in 2010 was preventable the assumption that the age pattern of vision impairment
or treatable compared with 80% (78–83) in 1990. in the study matched the modelled pattern in the country
Our findings are in agreement with those of previous where the study was done.12 Third, the definition of the
studies that assessed the causes for vision loss within diseases varied between studies. For instance, some
countries, regions, or worldwide. The combined data studies defined glaucoma according to the criteria of the
from studies done in Australia show that age-related International Society for Geographical and
macular degeneration was the major cause of severe Epidemiological Society,23 whereas in others the
vision impairment or blindness (defined as visual acuity appearance of the optic-nerve head was the main criterion,
in the better eye of less than 2/200), affecting 0·45% of independent of the presence of visual-field defects. In
the population.2 In a study in the Netherlands, myopic terms of glaucoma, differences in diagnostic criteria have
degeneration and optic neuropathy were the main causes led to a 12-fold difference in reported prevalence.24 Fourth,
of impaired vision among people younger than 75 years, the classification for macular degeneration included any
whereas among people aged 75 years or older, age-related macular disease and, therefore, we could not necessarily
macular degeneration was the major cause of blindness.3 differentiate between age-related macular degeneration,
In the Beijing Eye Study,18 the most frequent causes of myopic maculopathy, and other retinal or macular
blindness or MSVI were cataract, degenerative myopia, disorders. Since myopic retinopathy is ranked among the
glaucoma, corneal opacity, and other optic-nerve damage. three most common causes of vision impairment and
Age-related macular degeneration and diabetic blindness in some studies from east Asia, this mixed

www.thelancet.com/lancetgh Vol 1 December 2013 e347

 Articles

grouping might prove important. The data from this important for the setting of priorities, development of
analysis, therefore, cannot be used to reflect the role of policies, and for planning (panel). Additionally, our data
myopic retinopathy, including myopic maculopathy, in provide a resource for advocacy efforts to mobilise
vision impairment. Fifth, unidentified or other causes resources for eye-care services from governments,
generally represented 20–30% of cause for MSVI and donors, and civil society.
20–35% for blindness. Sixth, for 20 countries that are Contributors
deemed by WHO to be endemic for trachoma, data on the RRAB, GAS, JLS, and HP prepared the vision impairment survey data.
prevalence of trachoma or trichiasis were not available. GAS, RAW, JLS, and SRF analysed the data. RRAB, GAS, and JBJ wrote
the first draft of the report. All authors contributed to the study design,
We had, therefore, to assume a conservative proportion of analysis, and writing of the report. RRAB oversaw the research.
zero. This approach could have led to an underestimation
Conflicts of interest
of the prevalence of trachoma as cause for blindness and We declare that we have no conflicts of interest.
MSVI. Seventh, protocol dictates that population-based
Acknowledgments
studies will report one principal cause per an individual This study was funded by the Bill & Melinda Gates Foundation, Fight for
assessed to arrive at the causal prevalence. When Sight, the Fred Hollows Foundation, and the Brien Holden Vision
individuals had multiple disorders that could have Institute. GAS is a staff member of WHO. The author alone is
contributed equally to visual loss, only the one deemed responsible for the views expressed in this publication and they do not
necessarily represent the decisions, policy, or views of WHO. A list of
most readily curable or most easily preventable was members of the Vision Loss Expert Group appears at http://www.anglia.
recorded.25 This approach has the potential to ac.uk/ruskin/en/home/microsites/veru/other_research_areas/global_
underestimate the contribution of diabetic retinopathy, burden_of_diseases.html.
glaucoma, or other diseases with cataracts at presentation References
and to underestimate the burden of cataract when patients 1 Sommer A, Tielsch JM, Katz J, et al. Racial differences in the
cause-specific prevalence of blindness in east Baltimore.
also have an uncorrected refractive error.26 Finally, some N Engl J Med 1991; 325: 1412–17.
studies have small sample sizes, which leads to large 2 Taylor HR, Keeffe JE, Vu HTV, et al. Vision loss in Australia.
confidence intervals for cause-specific prevalence Med J Aust 2005; 182: 565–68.
estimates. Our methods take into account sample size, 3 Klaver CC, Wolfs RC, Vingerling JR, Hofman A, de Jong PT.
Age-specific prevalence and causes of blindness and visual
which should lessen the effect of small studies on the impairment in an older population: the Rotterdam Study.
estimates compared with that of large studies. Arch Ophthalmol 1998; 116: 653–58.
The strengths of our study include the amount of 4 Congdon N, O’Colmain B, Klaver CC, et al, for the Eye Diseases
Prevalence Research Group. Causes and prevalence of visual
population-based data accessed and used; analysis of impairment among adults in the United States. Arch Ophthalmol
trends in the causes of vision impairment; incorporation 2004; 122: 477–85.
of non-linear age trends and the taking account of data 5 Thylefors B, Négrel AD, Pararajasegaram R, Dadzie KY. Global data
on blindness. Bull World Health Organ 1995; 73: 115–21.
that were not reported by age; and systematic quantitative 6 WHO. Global initiative for the elimination of avoidable blindness
analysis and reporting of uncertainty. The large network (WHO/PBL/97.61 Rev 2). Geneva: World Health Organization,
of ophthalmological researchers involved in identification 2000.
7 Resnikoff S, Pascolini D, Etya’ale D, et al. Global data on visual
and assessment of data sources ensured that we were impairment in the year 2002. Bull World Health Organ 2004;
able to access unpublished as well as published data: 82: 844–51.
unpublished data were from 48 population-based studies, 8 Pascolini D, Mariotti SP. Global estimates of visual impairment:
four government reports, and 44 rapid assessment of 2010. Br J Ophthalmol 2012; 96: 614–18.
9 Murray CJ, Vos T, Lozano R, et al. Disability-adjusted life-years
cataract surgical services and rapid assessment of (DALYs) for 291 diseases and injuries in 21 regions, 1990–2010:
avoidable blindness surveys. These data meant that a systematic analysis for the Global Burden of Disease Study 2010.
studies for which investigators had published only Lancet 2013; 380: 2197–223.
10 Vos T, Flaxman AD, Naghavi M, et al. Years lived with disability
summary data could be included. Thus, we assessed all (YLDs) for 1160 sequelae of 289 diseases and injuries 1990–2010:
major studies of vision impairment and included only a systematic analysis for the Global Burden of Disease Study 2010.
those studies that met our inclusion criteria for Lancet 2013; 380: 2163–96.
11 Bourne R, Price H, Taylor H, et al, for the Global Burden of Disease
representativeness of the population and clarity of visual Vision Loss Expert Group. New systematic review methodology for
acuity procedures and definitions. visual impairment and blindness for the 2010 Global Burden of
In conclusion, cataract or uncorrected refractive error Disease study. Ophthalmic Epidemiol 2013; 20: 33–39.
12 Stevens G, White R, Flaxman SR, et al. Global prevalence of visual
led to 54% of blindness and 71% of MSVI in 2010. Both impairment and blindness: magnitude and temporal trends,
causes are easily and completely treatable. Glaucoma, 1990–2010. Ophthalmology 2013; published online July 10.
macular degeneration, and trachoma were less frequent DOI:10.1016/j.ophtha.2013.05.025.
causes of vision loss. The proportions of vision 13 Smith J, Mann R, Haddad D, et al. Global Atlas of Trachoma. http://
www.trachomaatlas.org (accessed Oct 17, 2013).
impairment caused by cataract and trachoma decreased 14 Peterson HM, Flaxman AD. Meta-regression with DisMod-MR: how
in the study period of 20 years; those for glaucoma, robust is the model? Lancet 2013; 381: S110.
macular degeneration, and diabetic retinopathy 15 Myerson R, Rosenfeld LC, Lim SS, Murray CJL. Safe pregnancy and
delivery: a systematic analysis of the trends in the coverage of
increased; and that for uncorrected refractive error antenatal and intrapartum care. In: Horton R, ed. Global health
changed little. These temporal changes in cause-specific metrics and evaluation: controversies, innovation, accountability.
prevalence of blindness and vision impairment are Seattle, WA: Lancet, 2011: 85.

e348 www.thelancet.com/lancetgh Vol 1 December 2013

 Articles

16 WHO Global Health Observatory Data Repository. Trachoma: status 22 Nangia V, Jonas JB, Sinha A, Gupta R, Bhojwani K. Prevalence of
of endemicity for blinding trachoma by country. http://apps.who. undercorrection of refractive error in rural Central India. The
int/gho/data/node.main.A1645?lang=en (accessed Oct 17, 2013). Central India Eye and Medical Study. Acta Ophthalmol 2012;
17 Ahmad O, Boschi-Pinto C, Lopez AD, et al. Age standardization of 90: E166–67.
rates: a new WHO standard. Geneva: World Health Organization, 23 Foster PJ, Buhrmann R, Quigley HA, et al. The definition and
2001. classification of glaucoma in prevalence surveys. Br J Ophthalmol
18 Xu L, Wang Y, Li Y, et al. Causes of blindness and visual impairment 2002; 86: 238–42.
in an urban and rural area in Beijing: the Beijing Eye Study. 24 Wolfs RC, Borger PH, Ramrattan RS, et al. Changing views on
Ophthalmology 2006; 113: 1141.e1–e11. open-angle glaucoma: definitions and prevalences—The Rotterdam
19 Schellini SA, Durkin SR, Hoyama E, et al. Prevalence and causes of Study. Invest Ophthalmol Vis Sci 2000; 41: 3309–21.
visual impairment in a Brazilian population: the Botucatu Eye 25 WHO. Coding instructions for the WHO/PBL eye examination
Study. BMC Ophthalmol 2009; 9: 8. record (version III). http://www.who.int/ncd/vision2020_
20 Chang C, Cañizares R, Cuenca VJ, et al. Investigación rápida de la actionplan/ documents/pbl_88_1.pdf (accessed Oct 17, 2013).
ceguera evitable. Estudio RAAB—Ecuador. Quito: RM Soluciones 26 Kempen JH. The need for a revised approach to epidemiological
Grafica’s, 2009. monitoring of the prevalence of visual impairment.
21 Xu L, Li J, Cui T, Jonas JB. Frequency of under-corrected refractive Ophthalmic Epidemiol 2011; 18: 99–102
error in elderly Chinese in Beijing.
Graefes Arch Clin Exp Ophthalmol 2006; 244: 871–73.

www.thelancet.com/lancetgh Vol 1 December 2013 e349