Aseries of efforts, such as including cataracts in most national plans for the prevention of visual impairment^[1], have been made to improve cataract-related health services. However, cataracts remain a major public health problem^[2]. In 2010, cataracts were responsible for 33.4% of global blindness and 18.4% of global moderate to severe vision impairment^[2]. With the increasing life expectancy and rapidly aging population, the number of people with vision impairment due to cataracts is expected to increase continuously. The burden of cataracts can be quantified by disability adjusted life year (DALY) as the sum of year of life lost (YLL) and year lived with disability (YLD) for each location estimated by the Global Burden of Diseases (GBD), Injuries, and Risk Factors Study 2017^[3]. The GBD study 2017 was the result of a global collaboration to examine data on 359 diseases and injuries in 195 countries and territories, and it revealed that the global DALYs for cataracts increased by 29.6% from 2007 to 2017, reaching 8,010 thousands^[3].

Cataracts are the most unevenly distributed noncommunicable eye disease in the world, placing the greatest burden on middle- and low-income countries^[4], which result from the combined effects of socioeconomic and environmental factors. From a socioeconomic perspective, previous studies have found that the health burden of cataract vision loss is correlated with socioeconomic status^{[5, 6]}. Cataracts can not only cause vision loss but also cause more serious blindness^{[1, 2]}, which leading to an increased risk of death^[7] and impaired quality of life^[8]. We speculate that burden of cataract-related blindness is also correlated with socioeconomic status. From an environmental perspective, as it is well known, exposure to ultraviolet radiation (UVR) is one of the risk factors for cataract blindness, and the association of a higher cataract blindness burden with UVR has been documented in numerous reports^[9-11]. A study concluded that the prevalence of cataracts increased at a rate of 3% for each degree of latitude to the south^[12]. Therefore, UVR is a major factor leading to an uneven worldwide distribution of the cataract-related blindness burden, and it cannot be ignored. However, little evidence of evaluation of the effect of co-exposure to high-UVR and poor socioeconomic status on burden of cataract-related blindness.

We hypothesized that co-exposure on high-UVR and poor socioeconomic development jointly leads to the inequality in global cataract-related burden, with a stronger association of socioeconomic status with cataract-related burden in high-UVR countries than in countries of low UVR exposure. Therefore, in addition to evaluating the association of socioeconomic status and the burden of cataract-related blindness, we further explored the potential interaction effect between socioeconomic status and UVR exposure on the burden of cataract blindness in 185 countries and territories worldwide. Our study may be informative and helpful in achieving the aim proposed by the World Health Organization (WHO) Global Action Plan (GAP)^[1] of eliminating blindness caused by cataracts.

This study involved all countries in the world, using both country-level and subnational-level data. In the country-level analysis, countries with data in the GBD study 2017 but not in the Human Development Report were excluded (Supplementary Table S1 available in www.besjournal.com). We also analyzed subnational-level data if subnational regions were included in the GBD study 2017 (Supplementary Table S2 available in www.besjournal.com). All variables used the latest values, except UVR exposure and gross domestic product (GDP) per capita (Supplementary Table S3 available in www.besjournal.com). Satellite-derived UVR exposure data were incomplete from June 1–14, 2016 and from May 12–16, 2017, so we used 2015 solar UVR data. The GDP per capita of Japan was available only in 2014.

Global Burden of Cataract Blindness　The GBD study 2017 provides YLD to quantify the burden of cataract blindness. Age-standardized YLD rates associated with blindness due to cataracts were analyzed. The data were derived from the open-access database of the Global Burden of Disease study 2017 (http://ghdx.healthdata.org/gbd-results-tool), which contains quantitative data on nonfatal health outcomes in terms of YLDs for a list of 354 GBD causes according to different severity splits in 195 countries^[13]. The YLDs were estimated as the product of a prevalence estimate and a disability weight for the health states of each mutually exclusive sequela, adjusted for comorbidity^[13]. Disability weights employed numbers on a scale from 0 to 1 that represented the severity of health loss associated with a single given health state. Regarding cataracts (International Classification of Diseases 10^th Revision (ICD-10) codes H25-H26 and H28-H28.8), the disability weight was 0.187 (0.124–0.260) for blindness^[13]. YLD rates were calculated by dividing the number of YLDs by the relevant population. The GBD study 2017 reference population was used to calculate the age-standardized YLD rate^[14]. The following GBD study 2017 data concerning cataract blindness were collected as the outcome variables: (1) national age-standardized YLD rates owing to cataract blindness in 2017 and (2) subnational age-standardized YLD rates owing to cataract blindness in 2017. Because the GBD study data can be downloaded from an open access database, ethics approval and informed consent were not required for this study.

Human Development Index　The human development index (HDI), as a regional socioeconomic indicator, is a composite measure of health, education, and income, measured by life expectancy at birth, mean years of schooling and gross national income per capita, respectively^[15]. A higher HDI value indicates a higher level of socioeconomic development, ranging from 0 to 1. Country-level HDI data were obtained from the Human Development Report 2018 released by the United Nations Development Programme (UNDP) (http://hdr.undp.org/en/data). Subnational-level HDI data were obtained from The Global Data Lab (https://hdi.globaldatalab.org/areadata/shdi/) of the Institute for Management Research, Radboud University. Using the UNDP categorization^[15], countries and subnational regions were divided into four socioeconomic groups: the low-HDI group (less than 0.550), medium-HDI group (0.550–0.699), high-HDI group (0.700–0.799) and very-high-HDI group (0.800 or greater).

Solar Ultraviolet Radiation Exposure　The estimated daily cloud-adjusted ambient solar UVR data were obtained from NASA Goddard Earth Sciences Data and Information Services Center readings from the Ozone Monitoring Instrument (OMI) mounted on the NASA Earth Observing System Auraspacecraft^[16]. The OMI is a nadir viewing spectrometer that measures solar reflected and backscattered radiation in the 270–500 nm wavelength range with a spectral resolution of approximately 0.5 nm in the UVR range. The OMI ultraviolet data consider the impact of altitude, ozone, surface albedo, aerosols and cloud coverage to accurately measure the amount of solar UVR that reaches the Earth’s surface^[17]. We estimated the UVR for analysis by averaging the daily estimates (in J/m²) in 2015 with an OMI Level 3 surface UV irradiance product, which was provided on a 1° × 1° (longitude × latitude) grid, with each cell covering an area of 110 km (north–south) × 66 km (east–west) according to the World Geodetic System 84 coordinate system. An average daily UVR level was obtained for each country using ArcGIS version 10.2 software (http://www.esri.com/software/arcgis/index.html). First, we built a raster layer in the ArcGIS program with UVR data. Second, we built a superimposed polygon vector layer on the basis of the raster layer with a world map of national and subnational borders. Third, we used the zonal statistics tool to quantify the UVR level per country. Subnational UVR data were calculated for limited countries with the corresponding value of the subnational burden of cataract blindness provided by the GBD study 2017 following similar procedures.

The covariates were selected based on the GBD study 2017 (Supplementary Table S4 available in www.besjournal.com) and previous literature^{[4, 5, 6]}. We further selected covariates if they showed a significant association (P < 0.05) with the burden of cataract blindness in the univariate analysis or if one of the regression coefficients changed by at least 10% after covariates were added to the multivariable-adjusted model. Overall, we controlled for the following potential confounding variables: country-specific male to female sex ratio, proportion of population using solid fuels, age-standardized prevalence of current tobacco smoking, age-standardized diabetes mellitus prevalence, proportion of population living in urban areas, population-mean body mass index (BMI), and GDP per capita at nominal values. Supplementary Table S3 in lists additional information for each confounder. Subnational-level covariates including male to female ratio, age-standardized diabetes mellitus prevalence and GDP per capita at nominal values were measured by region (Supplementary Table S5 available in www.besjournal.com). Other subnational variables were supposed to be homogeneous within each country.

Data are presented as the mean ± standard deviation (SD) or median (interquartile) for continuous variables and as frequency or percentage for categorical variables. Linear regression models were used to evaluate the associations between HDI, UVR exposure, and cataract age-standardized YLD rate owing to blindness using country-level data and subnational-level data in three steps. First, we examined conditions of normality and log-transformed the outcome measures if the normality assumption was violated. Second, we built an adjusted model depending on the inclusion of covariates. Third, interaction and stratified analyses were conducted according to UVR exposure (high UVR and low UVR), HDI status (low HDI, medium HDI, high HDI, and very high HDI), and the burden of cataract-related blindness. All analyses were performed with the statistical software packages R (http://www.R-project.org, The R Foundation) and EmpowerStats (http://www.empowerstats.com, X&Y Solutions, Inc., Boston, MA). A two-sided significance level of 0.05 was used to evaluate statistical significance.

The GBD study 2017 was based on a geographical hierarchy that included 195 countries and territories, of which 185 countries were also included in the Human Development Report. Both HDI and age-standardized YLD rates in 2017 were available for 185 countries (Table 1, Supplementary Table S1), covering 94.87% of all countries and territories worldwide (Supplementary Figure S1 available in www.besjournal.com), including 38 very-high-HDI, 39 high-HDI, 51 medium-HDI and 57 low-HDI countries. The global distribution of country-specific age-standardized YLD rates owing to blindness in 2017 was unequal (Supplementary Figure S2 available in www.besjournal.com). The geometric means of age-standardized YLD rates owing to blindness in each HDI group ranked from low to very high HDI, were as follows: 77.98, 45.47, 19.51, and 10.48. The subnational estimation of age-standardized YLD rates owing to blindness in the GBD study 2017 included 206 subnational regions belonging to seven countries: Japan, the United States, Sweden, the United Kingdom, Mexico, Brazil, and Indonesia (Supplementary Table S5). Of these subnational regions, 22 were in the medium-HDI group, 63 were in the high-HDI group, and 121 were in the very-high-HDI group, and the corresponding geometric mean of age-standardized YLD rates owing to blindness were 122.28, 51.28, and 4.64, respectively. Additional characteristics of the covariates in this study, stratified by the HDI of included countries and subnational regions, are shown in Table 1 and Supplementary Table S6 (available in www.besjournal.com). The mean country-specific daily UVR levels ranged from 732.19 to 4876.66 J·m⁻²·day⁻¹ (Supplementary Figure S3 available in www.besjournal.com). The mean daily UVR exposure doses declined from 3,696.64 (low HDI), 3,529.58 (medium HDI), 3,133.49 (high HDI), to 1,858.61 J·m⁻²·day⁻¹ (very high HDI) with the increase in HDI at the country level, and the trend was also observed at the subnational level, with a range from 4,430.78 (medium HDI) and 3,835.03 (high HDI) to 1,731.55 J·m⁻²·day⁻¹ (very high HDI).

Figure S1.  Global map of the HDI in countries included in the GBD study 2017. HDI, human development index; GBD, Global Burden of Disease

Figure S2.  Global map of health burden of cataract-related blindness with age-standardized YLD rates in 2017. YLD, year lived with disability

Figure S3.  Global map of UVR levels in 2015. UVR, ultraviolet radiation

National and subnational age-standardized YLD rates were log-transformed because of violations of the assumption of normality. In univariate analyses, population using solid fuels, current tobacco smoking prevalence, urbanization, BMI and GDP were significantly associated with the age-standardized YLD rate (P < 0.001) at the country level, and all covariates were significantly associated with the age-standardized YLD rate (P < 0.001) at the subnational level (Supplementary Table S7 available in www.besjournal.com). Urbanization and population using solid fuels were excluded due to collinear relationships in the subnational analysis.

Linear regression analysis showed that HDI was negatively associated with cataract age-standardized YLD rate owing to blindness in the crude model. In multivariable analyses, a consistent reverse association between HDI and the burden of cataract-related blindness among countries was retained in the adjusted model (Table 2). The adjusted model was adjusted for male to female sex ratio, GDP, population using solid fuels, age-standardized prevalence of current tobacco smoking, age-standardized diabetes mellitus prevalence, population living in urban areas and BMI mean. Very-high-HDI countries had an 84% lower age-standardized YLD rate [95% confidence interval (CI): 60%–93%, P < 0.001] compared to low-HDI countries. For high-HDI countries, the proportion was 76% (95% CI: 53%–88%, P < 0.001), and for medium-HDI countries, the proportion was 48% (95% CI: 15%–68%, P = 0.010; P for trend < 0.001) (Table 2). Although there was a lack of low-HDI subnational regions, the association between HDI and the burden of cataract-related blindness in subnational-level analyses had a similar trend, but a lower magnitude than that in the country-level analyses.

To assess potential effect modification by UVR exposure, we stratified the analysis by the median value of countries’ UVR exposure (high UVR > 3251.68 and low UVR ≤ 3251.68). The age-standardized YLD rate declined with increasing HDI levels in both UVR groups, and only countries with high HDIs in different UVR categories had significant differences (P < 0.05) (Figure 1). The mean UVR exposure of the low-UVR group decreased by the order of 2947.84 (low HDI), 2796.09 (medium HDI), 2073.29 (high HDI) and 1621.24 (very high HDI). Equivalent figures for the high-UVR group were 3810.10, 3940.34, 4004.37, and 3876.17, respectively. As shown in Figure 2 and Supplementary Figure S4 (available in www.besjournal.com), adjusted linear regression analysis indicated that the age-standardized YLD rate owing to cataract blindness was negatively correlated with HDI in both UVR categories in countries and subnational regions (P < 0.001). Table 3 presents the association of HDI with the cataract age-standardized YLD rate owing to blindness modified by UVR exposure in countries. UVR exposure was an effect modifier of HDI and the burden of cataract-related blindness in the adjusted model (P value for interaction = 0.047).

Figure 1.  Geometric mean of age-standardized YLD rate owing to cataract blindness across categories of socioeconomic status expressed as HDI at the country level. ^*P < 0.05

Figure 2.  Relationship between HDI and log age-standardized YLD rate owing to cataract blindness for both UVR categories at the country level after adjusting for all covariates. The lines represent fitted lines. HDI, human development index. YLD, year lived with disability. UVR, ultraviolet radiation.

Figure S4.  Relationship between HDI and log age-standardized YLD rate owing to blindness for both UVR categories at the subnational level after adjusting for all covariates. HDI, human development index; YLD, year lived with disability; UVR, ultraviolet radiation

This study showed that socioeconomic status was inversely correlated with the burden of cataract blindness and revealed that UVR exposure modified the association of socioeconomic status with the health burden of cataract blindness, taking into account differences in the male to female sex ratio, GDP, population using solid fuels, age-standardized prevalence of current tobacco smoking, age-standardized diabetes mellitus prevalence, population living in urban areas and BMI. Countries with higher levels of socioeconomic development were found to have lower cataract-related blindness burdens. When countries were replaced by subnational regions, the trend of the negative correlation of socioeconomic status with the burden of cataract blindness was largely retained but was lower in magnitude.

A previous study estimated the prevalence of and number of people with blindness due to cataracts and found that among 21 GBD study regions, the percentages of blindness caused by cataracts were lower in the high-income regions (< 15%) and higher (> 40%) in South and Southeast Asia and Oceania^[2]. Our study showed that the cataract-related blindness burden is more concentrated in countries with lower socioeconomic status. The HDI level was independently associated with the health burden of cataract-related blindness, with lower age-standardized YLD rates in higher HDI countries. A possible explanation is that the HDI level is related to the output and quality of cataract surgery. Cataract surgery is considered one of the most cost-effective health-care interventions, with a cost per DALY saved of US$ 20–40 million, and is performed with increasing frequency in all regions^[1]. However, barriers to cataract surgery services still exist in most countries; the most commonly cited barriers are socioeconomic factors, including income, insurance coverage and low government funding^[18-20]. A systematic review demonstrated inequalities in cataract surgery rates were found among countries grouped by income and were associated with socioeconomic indicators^[21]. Although the global initiative known as ‘VISION 2020: the Right to Sight’ has made many efforts to promote cataract surgery services at a cost that all patients can afford worldwide^[1], the cost still represents a significant expenditure, especially for patients of lower socioeconomic levels^[20]. In developing countries worldwide, over half of people with cataract blindness do not undergo cataract surgery^[22], mainly because of a low willingness to pay for it^[23]. In addition, the quality of cataract surgery is still a concern, with poorer outcomes related to low socioeconomic levels^[24]; this disparity aggravates between-country disparities in cataract blindness. Using subnational data, our study observed that the HDI was negatively correlated with the burden of cataract-related blindness. When subnational instead of national data were considered, the difference in socioeconomic status between regions within a country decreased, which helped us to observe the impact of socioeconomic status on the burden of cataract blindness at a smaller geographical coverage and at a detailed level. Although limited subnational data were included according to the GBD study, the findings suggest that socioeconomic disparities in the burden of cataract blindness exist not only among countries but also among subnational regions with different development levels.

Studies have shown that UVR exposure is one of the most important factors of cataractogenesis and is related to cataract development by inducing apoptosis and photooxidation^{[25, 26]}. Several findings have demonstrated that the association of UVR exposure with cataracts is dose dependent^{[27, 28]}. More UVR exposure implies an increasing risk of cataracts, leading to more cases of cataract blindness. The increased risk for cataract extraction in subjects exposed to high lifetime ambient total UVR (42.718 KJ·m⁻²) was confirmed [odds ratio (OR) = 1.53] by comparison with subjects exposed to moderate ambient total UVR (39.887 KJ·m⁻²)^[29]. High UVR exposure causes an increased cataract blindness burden and leads to added national medical expenditures related to socioeconomic level. The costs of environmental measures are often seen as an impediment to economic development. In regard to cataracts, the economic burden attributed to excess UVR is US$4.5 billion in the United States^[30]. The increase in economic costs due to high UVR may be unaffordable and present a concern for low- and middle-income countries. Ambient levels of UVR have been increasing and may persist at elevated levels in the future because of continuous stratospheric ozone depletion, and the increase represents an excess cost to address additional cases of cataracts^[31]. This serves as a warning that the countries with high UVR exposure and poor economic development must pay particular attention to ensuring protectionin order to reduce the incidence of cataract blindness. At low latitudes, there are a large number of countries that have lower economic conditions and suffer from high UV exposure. It is unrealistic to change the economic situation of these countries in a short time. Therefore, more economical interventions to protect the eyes from UVR exposure, to make the public realize that UVR exposure is harmful to the eyes, and to raise public recognition of UVR exposure. The WHO provides the UV Index (UVI) to guide crowd behavior, increase the population's attention to UVR exposure, and strengthen self-protection. In many countries the UVI is reported along with the weather forecast in newspapers, on TV, and on the radio. The publicity of UVI should be enhanced, and the public should be reminded to take eye protection measures such as wearing hats and sunglasses to avoid high UVR exposure in countries with lower socioeconomic status. Our findings may have significance for public health, given that cataracts are easily treatable^[32], and strengthening UVR protection could be an cost-effective intervention for delaying cataract blindness.

Several potential limitations need to be considered. First, the UVR exposure dose derived from the OMI surface UVR product might be overestimated compared to ground-based spectral UVR measurements^[33]. The estimated cataract-related blindness burden in the GBD study 2017 may be inadequate due to limited data sources and possible selection bias resulting from the reliance on clinical data records^[13]. Second, our study might be subject to ecological fallacy and bias, because the use of aggregate country-level data did not provide information on individuals. Third, the linear regression analysis included subnational-level data of age-standardized YLD rates owing to blindness, UVR exposure, HDI, male to female sex ratio, age-standardized diabetes mellitus prevalence and GDP. The remaining variables, such as population using solid fuels, age-standardized prevalence of current tobacco smoking, population living in urban areas and BMI mean, were not available at the subnational level. We assumed that these variables had a homogeneous distribution throughout each country, which covered important subnational differences. Furthermore, the study was restricted to the locations included in the GBD study 2017, therefore, only 206 subnational regions belonging to seven countries were included. For other countries with large geographic latitude coverage, we were forced to use a single national age-standardized YLD rate to represent the burden of cataract-related blindness without detailed subnational data. The insufficient sample size, especially the lack of low-HDI subnational regions, may cause poor representativeness. Despite these limitations, ecological studies are an effective method to explore associations on a worldwide level, especially based upon openly available data^[34].

In conclusion, long-term high-UVR exposure amplified the association of poor socioeconomic status with the burden of cataract-related blindness. The findings highlight that in addition to existing efforts toward eliminating cataract blindness, UVR exposure protection interventions, such as wearing glasses, wearing a cap and a reduction of outdoor activity time, must be reinforced in developing regions with high-UVR exposure to achieve the global target proposed by the WHO GAP.

The authors declare no conflict of interest.

DENG Yan calculated the solar ultraviolet radiation exposure, analyzed the data, and wrote the original article. YANG Dan obtained data on the global burden of cataract blindness and other related data. YU Jia Ming and XU Jing Xian determined the covariates related to cataracts. HUA Hui, CHEN Ren Tong, and WANG Nan performed the covariate calculations, statistical analysis and graphing. OU Feng Rong helped revise the manuscript. LIU Ru Xi and WU Bo performed data processing. LIU Yang designed the experiment. All authors contributed to the writing and editing of the final paper.