Global burden of low vision and blindness due to age-related macular degeneration from 1990 to 2021 and projections for 2050

Abstract

Background

Age-related macular degeneration (AMD) is a leading cause of blindness and low vision worldwide. This study examines the global burden and trends in AMD-related low vision and blindness from 1990 to 2021, with projections through 2050.

Methods

Data were obtained from the 2021 Global Burden of Disease (GBD 2021) study, covering 204 countries and regions. Key metrics, including the prevalent case numbers, annual disability-adjusted life years (DALYs), age-standardized prevalence rates (ASPR), and age-standardized DALY rates (ASDALYR), specific to low vision and blindness due to AMD, were calculated per 100,000 population. Trend analysis used the estimated annual percentage change (EAPC) method, and K-means clustering identified regions with similar burdens and trends. Autoregressive Integrated Moving Average(ARIMA) and Exponential Smoothing(ES) models provided future projections.

Results

Globally, the total number of prevalent cases and DALYs has substantially increased. The number of prevalent cases of low vision and blindness due to AMD increased from 3,640,180 (95% UI: 3,037,098 − 4,353,902) in 1990 to 8,057,521 (95% UI: 6,705,284-9,823,238) in 2021. DALYs increased from 302,902 (95% UI: 206,475 − 421,952) in 1990 to 578,020 (95% UI: 401,241–797,570) in 2021. From 1990 to 2021, both the ASPR and ASDALYR for AMD-related low vision and blindness showed a downward trend. The ASPR was 94 (95% UI: 78.32-114.42) per 100,000 population, with an EAPC of -0.26 (95% CI: -0.31 to -0.22), and the ASDALYR was 6.78 (95% UI: 4.7–9.32) per 100,000 population, with an EAPC of -0.94 (95% CI: -1.01 to -0.88). The disease burden of AMD-related low vision and blindness increases with age, and the burden for female patients is slightly higher than for males. Regional stratification by the Socio-Demographic Index (SDI) shows that the burden of AMD-related low vision and blindness in areas with low SDI is higher than in areas with high SDI. From 1990 to 2021, notable increases in ASPR and ASDALYR were observed mainly in the southern and central regions of sub-Saharan Africa. Moreover, the increases in prevalence and DALYs vary by region, country, and level of socioeconomic development. The ARIMA model predicts that by 2050, the number of prevalent cases of low vision and blindness due to AMD will reach 13,880,610(95% CI: 9,805,575–17,955,645), and the DALYs will be 764,731(95% CI: 683,535–845,926). The ES model predicts that by 2050, the number of prevalent cases of AMD-related low vision and blindness will reach 9,323,124(95% CI: 5,222,474–13,423,774), and the DALYs will be 641,451 (95% CI: 383,588–899,318).

Conclusion

This study indicates that between 1990 and 2021, the global prevalent cases and DALYs caused by AMD-related low vision and blindness have increased over the past three decades, correlating with factors such as age, gender, socioeconomic status, and geographical location. Predictive models indicate that as the population ages, the number of patients with low vision and blindness due to AMD, along with associated DALYs, will continue to rise. By 2050, it is expected that over 9 million people worldwide will be affected by AMD-related vision loss, with women being particularly impacted. These findings can provide data support for public health planning, resource allocation, and the formulation of medical policies, ensuring an effective response to the challenges posed by the future increase in AMD-related low vision and blindness.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12889-024-21047-x.

Introduction

Age-related macular degeneration (AMD) is a degenerative eye disease affecting the macula, primarily impacting the central region of the retina [1]. It is the leading cause of irreversible vision loss in individuals over 55 worldwide, accounting for 9% of cases of legal blindness globally [2]. The condition is characterized by progressive degeneration of the macular region, resulting in blurred central vision, distorted vision, and potentially leading to a central blind spot [1]. As the disease progresses, dark or blank spots may appear in the center of one’s field of vision, severely affecting daily activities such as reading, driving, and recognizing faces [3, 4]. In 2020, it was estimated that AMD affected approximately 196 million people globally [5], with 1.85 million experiencing blindness due to AMD [6]. Additionally, AMD-related moderate to severe vision impairment impacted 6.23 million individuals worldwide [6].In the population aged 45 to 85, the combined prevalence of any AMD is approximately 8.69% [5]. AMD has become an increasingly severe public health issue within globally aging societies, profoundly affecting patients’ quality of life and posing considerable challenges to health systems worldwide [7].

The Global Burden of Disease (GBD) database is a comprehensive global health research project aimed at providing a full and systematic assessment of diseases, injuries, and risk factors [8]. Previous studies have utilized the GBD database to analyze the global burden of AMD, uncovering various epidemiological trends and influencing factors. For example, studies have identified correlations between the age-standardized DALY rate for AMD and factors such as body mass index, serum cholesterol levels, alcohol consumption, and urbanization rate [9]. In addition, research has shown that the disease burden of AMD has increased over time, particularly affecting women, and is linked to socioeconomic status, education level, and geographical disparities [10, 11]; Furthermore, regional differences have been highlighted, with lower DALY rates in Africa and the Eastern Mediterranean compared to other WHO regions, such as the Americas and Southeast Asia, and a negative correlation between DALY rates and the Human Development Index (HDI) [12]. Although previous studies have provided valuable insights into the global prevalence and burden of AMD, they have limitations in terms of data updates and future trend forecasting. Systematic reviews and meta-analyses have shown that the prevalence of AMD varies significantly by ethnicity and region, with predictions indicating a future increase in the number of affected individuals [5]. Additionally, analyses combining data from multiple sources have projected a significant rise in new AMD cases worldwide by 2050, highlighting the urgent need for effective prevention and control measures [13]. However, several research gaps remain. Notably, most predictive studies broadly address AMD, with limited focus on low vision and blindness specifically associated with the disease. Additionally, many existing studies rely on systematic reviews and meta-analyses and do not always incorporate the latest available data, which may affect the precision of recent trend evaluations and burden estimates. Predictive studies, particularly those using GBD data for forecasting, remain limited. Moreover, most projections are country-specific, which can restrict insights into broader cross-regional trends and interactions, underscoring the need for a more comprehensive global perspective.

Therefore, this study utilizes data and methods from the GBD study project to analyze the global burden of low vision and blindness due to AMD from 1990 to 2021 and applies Autoregressive Integrated Moving Average and Exponential Smoothing models to forecast the situation in 2050. Employing these time series models can better handle non-stationary data and capture long-term trends, thereby enhancing the accuracy of future trend predictions. This study aims to provide a comprehensive analysis of the epidemiological trends and future projections of low vision and blindness due to AMD, which will not only assist in identifying and addressing major health threats, optimizing disease prevention and control strategies, and reducing health inequities, but also facilitate the advance planning of medical resources, reducing socio-economic burdens, and promoting global health cooperation. Consequently, this will lead to an overall enhancement of the quality of life and health standards of the elderly population.

Methods

Overview

This study draws on data from the GBD 2021 study, provided by the Institute for Health Metrics and Evaluation (IHME). The GBD 2021 employs standardized methodologies to estimate the global burden of diseases, including incidence, prevalence, mortality, and disability-adjusted life years (DALYs) across 371 diseases and injuries in 204 countries and territories. It covers 288 causes of death and 88 risk factors, with estimates for 21 countries and territories at the subnational level. GBD data sources include scientific literature, household surveys, epidemiological surveillance, disease registries, clinical informatics, and other data sources [14].

Data collection

Age-related macular degeneration (AMD) is defined as age-related deterioration of the macula, the part of the retina responsible for central vision, leading to central vision loss. The GBD 2021 model was used to assess the global burden of low vision and blindness due to AMD. Low vision includes both moderate and severe visual loss. The criteria follow WHO’s ICD-11 standards, evaluating the “presenting” or “best-corrected” vision of the better-seeing eye. Moderate vision loss is defined as visual acuity from 6/60 to < 6/18, severe vision loss as 3/60 to < 6/60, and blindness as < 3/60 or a visual field of < 10°. This dataset, spanning from 1990 to 2021, includes key metrics such as (1) the number of prevalent cases, (2) age-standardized prevalence rates (ASPR, per 100,000 population), (3) DALYs, and (4) age-standardized DALY rates (ASDALYR, per 100,000 population). Data were stratified by five-year age groups, from ages 45–49 up to those aged 95 years and older.

Geographical classification

This study employs the Sociodemographic Index (SDI), which is based on income per capita, average years of education for individuals aged 15 and older, and fertility rates among women under 25. Countries were grouped into five SDI levels: low, lower-middle, middle, upper-middle, and high. Additionally, the GBD framework categorizes 204 countries and territories into 21 super-regions and 54 regions based on geographical proximity and epidemiological similarities. The framework also considers World Bank income levels, WHO regional divisions, and the strength of healthcare systems, classifying regions as advanced, basic, limited, or minimal. For further details, see Supplementary Table 9.

Estimation models and standardization

Prevalence and DALYs were estimated using DisMod-MR 2.1 (Disease Modelling Meta-Regression version 2.1), a Bayesian meta-regression tool that ensures consistency across estimates by accounting for variables such as age, sex, location, and time. To calculate DALYs, we combined years of life lost (YLL) and years lived with disability (YLD) using the formula: DALY = (Number of deaths × Standard life expectancy at age of death) + (Number of prevalent cases × Disability weight) [14]. Disability weights, which quantify health loss, were derived from surveys assessing functional impacts associated with different outcomes, including AMD-related impairments. For AMD-related low vision and blindness, the severity levels are categorized into moderate, severe, and blindness. Each severity level is assigned an independent disability weight. These disability weights are applied separately for each level of impairment and are used to calculate the YLD for each sequela. For nonfatal diseases, DALYs are based solely on YLD. Estimates were generated with 95% uncertainty intervals (UIs), indicated by the 2.5th and 97.5th percentiles from 500 draws. The study also incorporates Age-standardized Prevalence Rates (ASPR) and Age-standardized Disability-Adjusted Life Year Rates (ASDALYR), both standardized to the GBD reference population and expressed per 100,000 population, enabling comparisons across regions and time periods [14].

Statistical analysis

We conducted the following statistical analysis on the data: (1) Prevalence, DALY, and age-standardized rates calculation: Calculated the number of prevalent cases, ASPR, DALYs, and ASDALYR for each year. (2) Stratified analysis: Performed stratified analysis on the prevalence and DALY rates by age, gender, SDI, and geographical location. (3) Trend analysis: Analyzed the trend of change in age-standardized rates (ASR) from 1990 to 2021 using the estimated annual percentage change (EAPC) method through a log-linear regression model y = α + βx + ϵ (where y = ln(ASR) and x is the year). The EAPC was reported along with a 95% confidence interval (CI). (4) Cluster analysis: Grouped GBD regions with similar EAPC trends into clusters based on indicators such as prevalence and DALY rates using the K-means clustering method.

Additionally, two forecasting models were employed for AMD-related low vision and blindness projections.

Autoregressive integrated moving average (ARIMA)

The ARIMA model, represented as ARIMA(p, d,q), is well-suited for time series data with consistent trends. In this model, p represents the order of the autoregressive terms, d refers to the number of differences required for stationarity, and q denotes the order of the moving average terms. By adjusting these components, ARIMA can effectively handle complex data and optimize predictive accuracy. For datasets where past values inform future outcomes through identifiable trends, ARIMA proves particularly effective. Given that low vision and blindness due to AMD exhibit long-term trends driven by demographic changes, ARIMA’s ability to model autoregressive, differencing, and moving average components makes it especially suitable for this type of epidemiological analysis. The ARIMA model is applied across multiple datasets, including age-standardized prevalence, age-standardized DALYs, prevalent cases, and DALYs, as demonstrated in the following equations:

Wherein, and represent the autoregressive and moving average polynomials, respectively, B is the backshift operator, and is the white noise error term.

Exponential smoothing

ES models, particularly the Holt method, were chosen to complement ARIMA by focusing on smoothing short-term fluctuations and projecting long-term trends. The damped trend variant of ES is particularly useful in healthcare settings, as it assigns higher weights to more recent observations, allowing for the capture of gradual changes influenced by recent trends in treatment or prevention strategies. This method is effective for projecting steady trends into the future, with the damping factor set to 0.9, preventing overly optimistic or pessimistic forecasts and ensuring that long-term trends realistically decay over time. Additionally, ES is simple, robust, and computationally efficient, making it ideal for data with consistent patterns, particularly for non-seasonal time series. The Holt model was applied to four datasets, including age-standardized prevalence, age-standardized DALYs, prevalence cases, and DALYs. The predictive equations are as follows:

Here, denotes the observed value at time t, represents the smoothed estimate, is the trend component, α and 𝛽 are the smoothing parameters, and 𝜙 (set at 0.9) is the damping factor governing the rate of decay for the trend over time. The parameter h represents the forecast horizon, indicating how many periods into the future the model is projecting. The inclusion of the damping factor allows the model to modulate its responsiveness to anticipated trends, rendering it apt for projections over extended horizons.

Model evaluation

The accuracy of the ARIMA and ES models was evaluated using a combination of statistical metrics. Data from 2021 were retained as a validation set to evaluate the ARIMA and ES accuracy. Specifically, we employed:

Root Mean Square Error (RMSE): This metric measures the standard deviation of the residuals (prediction errors). A lower RMSE indicates a better fit of the model to the data. We calculated RMSE for both the training and testing datasets to ensure the model’s robustness.

Mean Absolute Percentage Error (MAPE): This metric expresses the average magnitude of the percentage errors as a percentage of the actual values. It provides a relative measure of prediction accuracy.

All statistical analyses and visualizations were conducted using R software (Version 4.0.3), with statistical significance set at a two-tailed P-value of < 0.05.

Results

Global trends

From 1990 to 2021, the global number of prevalent cases and DALYs for low vision and blindness due to AMD showed a notable increase, while the age-standardized rates declined. In 1990, there were 3,640,180 prevalent cases (95% UI: 3,037,098–4,353,902) with an ASPR of 99.5 (95% UI: 83.16–118.04) per 100,000 population. By 2021, prevalent cases rose to 8,057,521 (95% UI: 6,705,284–9,823,238), though ASPR slightly declined to 94 (95% UI: 78.32–114.42) per 100,000 population, with an EAPC of -0.26 (95% CI: -0.31 to -0.22) (Fig. 1A). Global DALYs increased from 302,902 (95% UI: 206,475–421,952) in 1990 to 578,020 (95% UI: 401,241–797,570) in 2021. However, the global ASDALYR declined from 8.38 (95% UI: 5.7–11.53) per 100,000 population in 1990 to 6.78 (95% UI: 4.7–9.32) per 100,000 population in 2021, with an EAPC of -0.94 (95% CI: -1.01 to -0.88) (Fig. 1B).

Fig. 1.

Global burden of low vision and blindness due to AMD from 1990 to 2021: age-standardized rates and case numbers. (A) Age-standardized prevalence rate and number of prevalent cases from 1990 to 2021. (B) Age-standardized DALYs rate and DALYs from 1990 to 2021

Burden of low vision and blindness due to AMD by age

In 2021, ASDALYR and ASPR increased substantially with age, as did the number of prevalent cases and DALYs. The 65–69 age group recorded the highest counts, with 1,554,182 prevalent cases (95% UI: 1,179,507–2,027,660) and 102,046 DALYs (95% UI: 67,067–147,824) (Fig. 2B; Tables 1 and 2).

Fig. 2.

Global burden of low vision and blindness due to AMD by age in 2021: age-standardized rates and case numbers. (A) Age-standardized rates of prevalence and DALY rates by age in 2021. (B) Number of prevalent cases and DALYs by age in 2021

Table 1.

The estimated number and age-standardized rates (per 100,00 population population) of low vision and blindness due to AMD DALYs and temporal trends by sex, age and SDI from 1990 to 2021

Characterstics	1990		2021		1990–2021
	Number(95%UI)	ASR(95%UI)	Number(95%UI)	ASR(95%UI)	EAPC(95%CI)
Global	302,902 (206,475 − 421,952)	8.38 (5.70–11.53)	578,020 (401,241–797,570)	6.78 (4.7–9.32)	-0.94 (-1.01–0.88)
Sex
Female	183,753 (125,264 − 254,818)	9.04 (6.16–12.45)	345,411 (239,405–473,686)	7.38 (5.11–10.12)	-0.94 (-1.01–0.87)
Male	119,149 (80,936 − 167,890)	7.37 (5.01–10.24)	232,609 (159,806 − 323,231)	6.01 (4.15–8.29)	-0.88 (-0.94–0.82)
Age
45–49 years	1,470 (722–2,766)	0.63 (0.31–1.19)	2,080 (1,027 − 3,837)	0.44 (0.22–0.81)	-1.43 (-1.5-- 1.36)
50–54 years	10,780 (5,735 − 17,207)	5.07 (2.7–8.09)	16,674 (9,382 − 26,801)	3.75 (2.11–6.02)	-1.27 (-1.35–1.18)
55–59 years	27,867 (17,299 − 41,617)	15.05 (9.34–22.47)	46,521 (29,033–69,918)	11.76 (7.34–17.67)	-1.07 (-1.14–1)
60–64 years	46,609 (29,640 − 69,823)	29.02 (18.45–43.47)	76,600 (50,171 − 113,578)	23.93 (15.68–35.49)	-0.95 (-1.05–0.85)
65–69 years	54,220 (35,811 − 78,790)	43.86 (28.97–63.74)	102,046 (67,067–147,824)	36.99 (24.31–53.59)	-0.83 (-0.9–0.75)
70–74 years	49,601 (33,483 − 69,510)	58.59 (39.55–82.1)	100,488 (68,191 − 142,406)	48.82 (33.13–69.18)	-0.82 (-0.9–0.74)
75–79 years	46,237 (31,409 − 65,102)	75.11 (51.03-105.76)	83,927 (57,879 − 117,635)	63.64 (43.89–89.2)	-0.86 (-0.94–0.77)
80–84 years	35,933 (23,833 − 51,281)	101.57 (67.37-144.96)	69,960 (46,758 − 98,844)	79.88 (53.39-112.86)	-0.98 (-1.04–0.91)
85–89 years	20,405 (13,275 − 28,668)	135.03 (87.85-189.71)	47,427 (31,173 − 66,496)	103.73 (68.18-145.44)	-1.07 (-1.14–1)
90–94 years	7,481 (4,973 − 10,626)	174.58 (116.05-247.97)	23,088 (15,434 − 32,979)	129.06 (86.27-184.35)	-1.11 (-1.17–1.05)
95 + years	2,300 (1,471–3,476)	225.87 (144.5-341.41)	9,210 (5,996 − 13,719)	168.99 (110.02-251.71)	-1.06 (-1.11–1.02)
SDI region
High-middle SDI	77,003 (52,263 − 105,769)	8.65 (5.88–11.82)	147,383 (102,643 − 202,092)	7.4 (5.15–10.14)	-0.77 (-0.86–0.68)
High SDI	61,276 (41,090 − 82,506)	5.48 (3.68–7.35)	95,839 (65,148 − 129,960)	4.08 (2.76–5.49)	-1.06 (-1.12–0.99)
Low-middle SDI	62,339 (42,173 − 88,167)	11.43 (7.79–16.04)	101,899 (70,143–143,939)	7.64 (5.28–10.75)	-1.48 (-1.6–1.36)
Low SDI	23,489 (16,019–32,691)	11.9 (8.19–16.58)	44,623 (30,566 − 61,772)	10.08 (6.91–13.86)	-0.65 (-0.71–0.6)
Middle SDI	78,554 (53,555 − 110,860)	8.65 (5.85–12.13)	187,897 (129,233–259,067)	7.26 (5-9.96)	-1.03 (-1.16–0.91)

UI: uncertainty intervals; CI: confidence interval; SDI: socio-demographic index; EAPC: estimated annual percentage change

Table 2.

The estimated number and age-standardized rates (per 100,00 population population) of low vision and blindness due to AMD prevalence and temporal trends by sex, age and SDI from 1990 to 2021

Characterstics	1990		2021		1990–2021
	Number(95%UI)	ASR(95%UI)	Number(95%UI)	ASR(95%UI)	EAPC(95%CI)
Global	3,640,180 (3,037,098 − 4,353,902)	99.5 (83.16-118.04)	8,057,521 (6,705,284–9,823,238)	94 (78.32-114.42)	-0.26 (-0.31–0.22)
Sex
Female	2,141,000 (1,787,369–2,536,951)	104.34 (87.5-123.1)	4,657,829 (3,881,913–5,650,808)	99.53 (82.98-120.72)	-0.25 (-0.3–0.21)
Male	1,499,180 (1,245,793–1,812,002)	92.23 (76.87-110.83)	3,399,691 (2,808,352–4,167,120)	87.1 (71.99-106.29)	-0.23 (-0.27–0.19)
Age
45–49 years	11,290 (6,646 − 16,939)	4.86 (2.86–7.29)	19,066 (11,319 − 28,904)	4.03 (2.39–6.1)	-0.73 (-0.77–0.69)
50–54 years	100,789 (70,851 − 135,487)	47.41 (33.33–63.74)	189,097 (129,066–255,850)	42.5 (29.01–57.5)	-0.46 (-0.51–0.42)
55–59 years	315,885(234,195–409,185)	170.56 (126.46-220.94)	635,566(457,265–851,423)	160.61 (115.55-215.15)	-0.27 (-0.31–0.23)
60–64 years	579,244(429,365–764,115)	360.65 (267.34-475.76)	1,121,462(807,561-1,540,566)	350.4 (252.32-481.35)	-0.18 (-0.25–0.11)
65–69 years	699,154(538,893–896,557)	565.62 (435.96-725.31)	1,554,182(1,179,507-2,027,660)	563.43 (427.6-735.08)	-0.08 (-0.15–0.02)
70–74 years	644,282(503,818–820,529)	761.01 (595.1-969.19)	1,498,299(1,161,980-1,953,673)	727.9 (564.51-949.13)	-0.14 (-0.21–0.08)
75–79 years	567,756(451,929 − 697,373)	922.35 (734.18-1132.92)	1,190,223(944,328-1,499,312)	902.47 (716.03-1136.84)	-0.22 (-0.28–0.16)
80–84 years	407,664(320,705 − 508,369)	1152.38 (906.56-1437.05)	917,306(710,148-1,166,097)	1047.36 (810.83-1331.42)	-0.37 (-0.42–0.33)
85–89 years	216,953(174,912 − 268,275)	1435.72 (1157.51-1775.35)	576,229(456,942 − 713,117)	1260.3 (999.4-1559.69)	-0.56 (-0.63–0.5)
90–94 years	75,076(59,746 − 96,255)	1751.98 (1394.25-2246.23)	259,489(205,887 − 331,388)	1450.52 (1150.89-1852.43)	-0.7 (-0.76–0.65)
95 + years	22,086(16,846 − 28,883)	2169.32 (1654.69-2837.02)	96,601(73,087–125,311)	1772.4 (1340.98-2299.15)	-0.74 (-0.78–0.71)
SDI region
High-middle SDI	917,446(772,003 − 1,089,793)	100.42 (84.93-118.77)	2,103,144(1,762,018 − 2,558,108)	104.55 (87.88–126.6)	0.12 (0.05–0.18)
High SDI	641,152(536,445–759,449)	56.98 (47.61–67.51)	1,132,042(951,466-1,351,423)	48.43 (40.55–57.77)	-0.57 (-0.64–0.51)
Low-middle SDI	750,883(613,961 − 907,402)	138.32 (114.13-165.93)	1,403,317(1,142,219-1,725,627)	104.67 (85.46-127.94)	-1.03 (-1.16–0.91)
Low SDI	286,203(236,098–342,481)	144.7 (120.56-171.58)	623,529(507,719–765,708)	139.92 (114.54-171.01)	-0.23 (-0.28–0.18)
Middle SDI	1,041,645(862,866-1,275,940)	114.58 (94.89-138.84)	2,790,576(2,301,858-3,415,653)	107.56 (88.68-131.24)	-0.33 (-0.39–0.26)

UI: uncertainty intervals; CI: confidence interval; SDI: socio-demographic index; EAPC: estimated annual percentage change

From 1990 to 2021, ASDALYR and ASPR generally trended downward across age groups (Fig. 3A). The most rapid ASPR decline occurred in those aged 95 and above, with an EAPC of -0.74 (95% CI: -0.78 to -0.71), while the 45–49 age group showed the fastest ASDALYR decrease, with an EAPC of -1.43 (95% CI: -1.5 to -1.36) (Tables 1 and 2).

Fig. 3.

Global burden of low vision and blindness due to AMD by age from 1990 to 2021: age-standardized rates and case numbers. (A) Age-standardized rates of prevalence and DALYs by age from 1990 to 2021. (B) Numbers of prevalent cases and DALYs by age from 1990 to 2021

Burden of low vision and blindness due to AMD by gender

From 1990 to 2021, the ASDALYR and ASPR for females were consistently higher than those for males globally, although both showed a general decline (Fig. 4, Supplementary Fig. 1). In 1990, the ASPR for females was 104.34 (95% UI: 87.5–123.1) per 100,000 population, decreasing to 99.53 (95% UI: 82.98–120.72) per 100,000 population in 2021, with an EAPC of -0.25% (95% CI: -0.30 to -0.21). For males, ASPR decreased from 92.23 (95% UI: 76.87–110.83) per 100,000 population in 1990 to 87.1 (95% UI: 71.99–106.29) per 100,000 population in 2021, with an EAPC of -0.23% (95% CI: -0.27 to -0.19).

Fig. 4.

Global burden of low vision and blindness due to AMD by gender from 1990 to 2021: age-standardized rates and case numbers. (A) Age-standardized rates of prevalence and DALYs by gender from 1990 to 2021. (B) Number of prevalent cases and DALYs by gender from 1990 to 2021

Similarly, the ASDALYR for females was 9.04 (95% UI: 6.16–12.45) per 100,000 population in 1990, declining to 7.38 (95% UI: 5.11–10.12) per 100,000 population by 2021, with an EAPC of -0.94% (95% CI: -1.01 to -0.87). For males, ASDALYR decreased from 7.37 (95% UI: 5.01–10.24) per 100,000 population in 1990 to 6.01 (95% UI: 4.15–8.29) per 100,000 population in 2021, with an EAPC of -0.88% (95% CI: -0.94 to -0.82) (Fig. 4A).

Despite these slightly lower rates in males, the actual number of cases and DALYs increased over time. Female prevalent cases rose from 2,141,000 in 1990 to 4,657,829 in 2021, while for males, cases increased from 1,499,180 to 3,399,691. Similarly, DALYs for females grew from 183,753 in 1990 to 345,411 in 2021, and for males, from 119,149 to 232,609 (Fig. 4B), indicating a more substantial increase in females.

It is worth noting that the data and Fig. 4 highlight notable changes starting in 2019, particularly among females, with a increase in the total number of prevalent cases, rising from 4,407,255.07 in 2019 to 4,657,829.46 in 2021. Although the ASPR shows a temporary increase from 99.5435 per 100,000 population in 2019 to 101.1485 per 100,000 population in 2020, they return to 99.5257 per 100,000 population in 2021. In contrast, the ASDALYR among females exhibit a consistent upward trend, increasing from 7.2280 per 100,000 population in 2019 to 7.3796 per 100,000 population in 2021, along with a rise in the total number of DALYs from 320,225.00 in 2019 to 345,411.05 in 2021.

Burden of low vision and blindness due to AMD by SDI

Between 1990 and 2021, the burden of low vision and blindness due to AMD was higher in low and lower-middle SDI regions compared to high SDI regions. ASPR and ASDALYR were lowest in high SDI regions, with ASPR declining from 56.98 (95% UI: 47.61–67.51) per 100,000 population in 1990 to 48.43 (95% UI: 40.55–57.77) per 100,000 population in 2021, with an EAPC of -0.57 (95% CI: -0.64 to -0.51). ASDALYR decreased from 5.48 (95% UI: 3.68–7.35) per 100,000 population in 1990 to 4.08 (95% UI: 2.76–5.49) per 100,000 population in 2021, with an EAPC of -1.06 (95% CI: -1.12 to -0.99).

In low SDI regions, ASPR and ASDALYR were highest, though both showed a yearly decline. ASPR decreased from 144.7 (95% UI: 120.56–171.58) per 100,000 population in 1990 to 139.92 (95% UI: 114.54–171.01) per 100,000 population in 2021, with an EAPC of -0.23 (95% CI: -0.28 to -0.18). ASDALYR declined from 11.9 (95% UI: 8.19–16.58) per 100,000 population in 1990 to 10.08 (95% UI: 6.91–13.86) per 100,000 population in 2021, with an EAPC of -0.65 (95% CI: -0.71 to -0.60) (Fig. 5, Supplementary Fig. 2, Tables 1 and 2).

Fig. 5.

Global burden of low vision and blindness due to AMD by SDI from 1990 to 2021: age-standardized rates and case numbers. (A) Age-standardized rates of prevalence and DALYs by SDI from 1990 to 2021. (B) Number of prevalent cases and DALYs by SDI from 1990 to 2021

From 1990 to 2021, distinct EAPC trends emerged in ASPR and ASDALYR across SDI regions. Low-middle SDI regions saw the most reduction in ASDALYR (-1.48, 95% CI: -1.6 to -1.36), while high SDI regions also experienced a notable decline (-1.06, 95% CI: -1.12 to -0.99). In ASPR, high-middle SDI regions had the largest increase (0.12, 95% CI: 0.05–0.18), whereas low-middle SDI regions showed the most substantial decrease (-1.03, 95% CI: -1.16 to -0.91).

Burden of low vision and blindness due to AMD by GBD region

In comparing four world regions between 1990 and 2021, Asia recorded the highest number of prevalent cases and DALYs for low vision and blindness due to AMD. The number of prevalent cases in Asia rose from 2,134,042 (95% UI: 1,762,431-2,607,398) in 1990 to 5,302,638 (95% UI: 4,399,214-6,477,250) in 2021, with corresponding DALYs increasing from 160,815 (95% UI: 110,030–226,529) to 350,590 (95% UI: 243,178–485,009). Africa led in age-standardized rates, with ASPR of 149.84 (95% UI: 124.88-176.64) per 100,000 population in 1990, decreasing slightly to 146.42 (95% UI: 119.82-177.63) per 100,000 population by 2021. Concurrently, the ASDALYR was 12.6 (8.72–17.68) per 100,000 population in 1990 and also experienced a slight decrease, reaching 11.01 (95% UI: 7.56–15.42) per 100,000 population by 2021.

Within the 21 super-regions, Western Sub-Saharan Africa showed the highest ASPR, at 226.01 (95% UI: 185.74-273.72) per 100,000 population in 1990, rising to 229.48 (95% UI: 185.45-281.21) per 100,000 population by 2021, with an EAPC of -0.2 (95% CI: -0.3 to -0.1). High-income Asia-Pacific regions showed the lowest ASPR at 23.69 (95% UI: 19.74–28.57) per 100,000 population in 1990, decreasing to 20.57 (95% UI: 17.16–24.97) per 100,000 population in 2021, with an EAPC of -0.43 (95% CI: -0.50 to -0.36). The Middle East and North Africa had the highest ASDALYR at 16.73 (95% UI: 11.36–23.55) per 100,000 population in 1990, decreasing to 14.04 (95% UI: 9.6-19.61) per 100,000 population by 2021. The Caribbean had the lowest ASDALYR, with 2.22 (95% UI: 1.44–3.10) per 100,000 population in 1990, declining to 1.78 (95% UI: 1.17–2.50) per 100,000 population by 2021. Across 21 super-regions, the most ASDALYR increase was in Southern Sub-Saharan Africa (EAPC 0.27, 95% CI: 0.13 to 0.42), while South Asia exhibited the steepest ASDALYR decline (EAPC − 1.81, 95% CI: -1.99 to -1.63). ASPR showed a similar trend, with Southern Sub-Saharan Africa experiencing the highest growth (EAPC 0.38, 95% CI: 0.31 to 0.46) and South Asia having the most substantial decline (EAPC − 1.29, 95% CI: -1.48 to -1.1).

Figure 6 shows the trends in ASDALYR and ASPR across 54 GBD regions from 1990 to 2021. Regions with a significant increase are indicated in blue, mainly including the southern and central regions of sub-Saharan Africa. Regions with a significant decrease are indicated in red, including East Asia and the Pacific, Africa, the Middle East, Eastern Europe, Central Europe, the Caribbean, and North America, where health conditions have improved. Regions indicated in green and purple show a minor increase or have remained relatively stable, as seen in Fig. 6 and Supplementary Fig. 3.

Fig. 6.

Global regional trends in health outcomes from 1990 to 2021

Burden of low vision and blindness due to AMD by 204 countries

Among 204 countries, Nepal had the highest ASPR for low vision and blindness due to AMD, increasing from 308.55 (95% UI: 246.57–381.73) per 100,000 population in 1990 to 399.22 (95% UI: 315.31–507.37) per 100,000 population in 2021, with an EAPC of 0.56 (95% CI: 0.43–0.69). Conversely, Barbados had the lowest ASPR, with 8.88 (95% UI: 6.96–11.43) per 100,000 population in 1990, slightly declining to 8.77 (95% UI: 6.8–11.4) per 100,000 population in 2021, yielding an EAPC of 0.02 (95% CI: 0–0.05).

China reported the highest number of prevalent cases, rising from 882,596 (95% UI: 723,350–1,091,239) in 1990 to 2,647,261 (95% UI: 2,185,651–3,258,896) in 2021, with an EAPC of -0.04 (95% CI: -0.19–0.10). Niue and Tokelau had the fewest cases, with only one case each in both 1990 and 2021 (Supplementary Tables 4 and Supplementary Fig. 4).

Iran led in ASDALYR, which decreased from 30.07 (95% UI: 20.52–41.87) per 100,000 population in 1990 to 25.02 (95% UI: 17.16–34.88) per 100,000 population in 2021. Barbados had the lowest ASDALYR, with a decline from 0.6 (95% UI: 0.4–0.87) per 100,000 population in 1990 to 0.53 (95% UI: 0.35–0.77) per 100,000 population in 2021. China also recorded the highest number of DALYs, which increased from 53,960 (95% UI: 37,210–75,602) in 1990 to 153,220 (95% UI: 105,756–212,586) in 2021. Nauru and Niue had the lowest DALYs, with values close to zero in both 1990 and 2021 (Supplementary Tables 2 and Supplementary Fig. 4).

Between 1990 and 2021, the EAPC for ASDALYR and ASPR for most countries and regions ranged between − 1% and 1%. Significant ASPR increases were observed in Benin, Chad, Côte d’Ivoire, Gambia, and Niger (EAPC > 1), with Niger showing the highest increase at 1.88 (95% CI: 1.64–2.13). In contrast, Iceland, India, Italy, Malaysia, and South Korea experienced significant ASPR decreases (EAPC < -1), with Iceland showing the steepest decline at -1.9 (95% CI: -2.1 to -1.7) (Fig. 7A). For ASDALYR, notable decreases occurred in India, Malaysia, and Thailand (EAPC < -2), with Malaysia achieving the largest reduction at -2.68 (95% CI: -2.88 to -2.48). In contrast, ASDALYR increases were noted in Benin, Côte d’Ivoire, and Niger, with Benin leading at an EAPC of 1.43 (95% CI: 1.27–1.58) (Fig. 7B).

Fig. 7.

Global burden of low vision and blindness due to AMD by 204 countries from 1990 to 2021: age-standardized rates and case numbers. (A) EAPC for age-standardized prevalence rate. (B) EAPC for age-standardized DALY rate. (C) Changes in the number of prevalent cases. (D) Changes in the number of DALYs

The number of prevalent cases and DALYs increased to varying degrees across most regions worldwide, with some countries experiencing substantial increases (Fig. 7C and D).

ARIMA model projections for the global burden of low vision and blindness due to AMD: 2022–2050

The forecast analysis based on the ARIMA model predicts that from 2022 to 2050, the ASPR for low vision and blindness due to AMD will remain stable at 94 per 100,000 population, while the ASDALYR will hold steady at 6.77 per 100,000 population (Fig. 8A and B). The projected ASPR consistently remains at 94.00, with 95% confidence intervals suggesting minor variability around this estimate. For example, in 2022, the confidence interval ranges from 92.58 to 95.41, reflecting a high degree of precision. Although the central ASPR value stays constant over the forecast period, the confidence intervals gradually widen, reaching a range of 86.38 to 101.62 by 2050. In contrast, the ASDALYR is projected to remain constant at 6.77 per 100,000 population, but the confidence intervals show broader variability, starting from 6.28 to 7.27 in 2022 and expanding to 4.10 to 9.44 by 2050.

Fig. 8.

ARIMA model projections for global burden of low vision and blindness due to AMD 2022–2050. (A) Future trends in age-standardized prevalence rate. (B) Future trends in age-standardized DALYs rate. (C) Projected increase in total prevalent cases. (D) Projected increase in total DALYs. Note The red shaded area represents the 95% confidence interval (CI) of the projections, while the red dashed line indicates the forecasted trend. The widening of the CI over time reflects increasing uncertainty

Despite stable rates, the number of prevalent cases and DALYs is projected to increase substantially. Prevalent cases are expected to grow from 8,258,317 (95% CI: 8,133,550–8,383,084) in 2022 to 13,880,610 (95% CI: 9,805,575–17,955,645) in 2050, reflecting a yearly increase (Fig. 8C). Likewise, DALYs are predicted to rise from 556,283 (95% CI: 505,919–606,647) in 2022 to 764,731 (95% CI: 683,535–845,926) by 2050, indicating a similar upward trend (Fig. 8D; Supplementary Table 5).

Between 2022 and 2050, ASPR and ASDALYR trends for low vision and blindness due to AMD vary between genders. For males, ASPR remains at 87.10 per 100,000 population, with prevalent cases rising from 3,481,958 to 5,785,419. Male ASDALYR declines slightly from 5.95 to 5.71 per 100,000 population, while DALYs increase from 236,953 to 358,588. Among females, ASPR decreases slightly from 98.67 to 97.73 per 100,000 population, as prevalent cases grow from 4,813,010 to 7,912,360. Female ASDALYR also sees a slight decrease, from 7.38 to 7.23 per 100,000 population, with DALYs rising from 356,283 to 660,684 (Fig. 9; Supplementary Table 6).

Fig. 9.

ARIMA model projections of global low vision and blindness due to AMD from 2022 to 2050 by gender. (A) Future trends in age-standardized prevalence rate and prevalent cases by gender. (B) Future trends in age-standardized DALYs rate and DALYs by gender

Exponential smoothing model projections for the global burden of low vision and blindness due to AMD: 2022–2050

According to the ES model forecast, from 2022 to 2050, the ASPR of global low vision and blindness due to AMD is expected to decrease slightly from 93.87 (95% CI: 92.38–95.37) to 92.80 (95% CI: 72.39–113.21) per 100,000 population (Fig. 10A). The confidence intervals reflect the uncertainty inherent in these projections; for example, the 2022 interval (92.38 to 95.37) indicates a relatively narrow margin of uncertainty, while the broader interval for 2050 (72.39 to 113.21) suggests greater variability. Simultaneously, the ASDALYR are projected to decline modestly from 6.78 (95% CI: 6.26–7.30) to 6.75 (95% CI: 4.64–8.87) per 100,000 population (Fig. 10B). While the 2022 interval indicates a relatively precise estimate, the 2050 interval highlights a broader range, underscoring increased uncertainty in the long-term projection.

Fig. 10.

ARIMA model projections for global burden of low vision and blindness due to AMD: 2022–2050. (A) Future trends in age-standardized prevalence rate. (B) Future trends in age-standardized DALYs rate. (C) Projected increase in total prevalent cases. (D) Projected increase in total DALYs. Note The red shaded area represents the 95% confidence interval (CI) of the projections, while the red dashed line indicates the forecasted trend. The widening of the CI over time reflects increasing uncertainty

Despite these reductions in rates, both the number of prevalent cases and DALYs are expected to increase. Specifically, the number of prevalent cases is projected to grow from 8,255,476 (95% CI: 8,112,826–8,398,127) in 2022 to 9,323,124 (95% CI: 5,222,474–13,423,774) by 2050, exhibiting an annual upward trend (Fig. 10C). Similarly, the DALY count is expected to rise from 557,526(95% CI: 504,038–611,014) in 2022 to 641,451 (95% CI: 383,58–899,318) by 2050, with a continuous year-on-year increase (Supplementary Tables 7 and Fig. 10D).

Further breakdown by gender from 2022 to 2050 reveals that the ASPR for males will slightly decline from 87.17 to 87.11 per 100,000 population while the number of prevalent cases increases from 3,507,506 to 3,979,632. Simultaneously, the ASDALYR for males is expected to decrease from 5.99 to 5.77 per 100,000 population, with DALYs rising from 236,520 to 269,872. For females, ASPR is anticipated to decrease more markedly, from 99.07 to 91.39 per 100,00 population, as prevalent cases increase from 4,750,767 to 5,366,945. ASDALYR for females will also show a notable decrease, from 7.42 to 7.74 per 100,00 population, with the DALY count rising from 355,196 to 438,653 (Supplementary Tables 8 and Fig. 11).

Fig. 11.

ES model projections of global low vision and blindness due to AMD from 2022 to 2050 by gender. (A) Future trends in age-standardized prevalence rate and prevalent cases by gender. (B) Future trends in age-standardized DALYs rate and DALYs by gender

Validation of the global burden of low vision and blindness due to AMD based on 1990–2020 GBD data

Using the 2021 data as a validation set, we compared the predictions from both ARIMA and ES models based on data from 1990 to 2020. Our analysis shows that the actual ASPR and ASDALYR in 2021 were slightly lower than predicted by models (Table 3). Specifically, the 2021 ASPR of 94 per 100,000 population was slightly lower than the predictions from the 1990–2020 ARIMA (99.07) and ES models (97.80). The actual ASDALYR for 2021 was recorded at 6.78 per 100,000 population, which is quite close to the projections made by both the ARIMA and ES models.

Table 3.

Comparison of predicted and actual 2021 global burden of low vision and blindness due to AMD using ARIMA and ES models

Year	Age-standardized prevalence rate	Numer of prevalent cases	Age-standardized DALYs rate	Numer of DALYs cases
2021	94.00	8,057,521	6.78	578,020
ARIMA model
Predict Based on Data from 1990 to 2020	99.07	8,511,716	6.77	538,579
RMSE	5.068723	454195.2	0.01060671	39441.15
MAPE	0.05392339	0.0563691	0.001566417	0.07323184
ES model
Predict Based on Data from 1990 to 2020	97.80	8,381,679	6.80	542,223
RMSE	3.796422	324158.2	0.01680284	35797.08
MAPE	0.04038807	0.04023052	0.002471468	0.06601908

Both models were evaluated using RMSE and MAPE, with the ES model demonstrating lower errors, indicating potentially higher accuracy in its predictions. The ARIMA model showed a RMSE of 5.069 for the ASPR and 0.0106 for the ASDALYR, with MAPE of 5.392% and 0.157%, respectively. The RMSE for the number of prevalent cases was 454195.2, and for the number of DALY cases, it was 39441.15, with MAPE values of 5.637% and 7.323%, respectively. The ES model presented an RMSE of 3.796 for ASPR and 0.016 for ASDALYR, with MAPE values of 4.038% and 0.247%, respectively. The RMSE for the number of prevalent cases was 324158.2, and for the number of DALY cases, it was 35797.08, with MAPE values of 4.023% and 6.602%, respectively. Overall, the general consistency between the model predictions and the actual 2021 data indicates that both the ARIMA and ES models have shown a reasonable level of predictive accuracy.

Discussion

This study provides a detailed analysis of the global disease burden trends for low vision and blindness due to AMD from 1990 to 2021 and predicts future trends using two models. The results show that although the total number of prevalent cases and DALYs for low vision and blindness due to AMD worldwide have increased during this period, the global ASDALYR and ASPR have declined. This phenomenon may be attributed to improvements in diagnostic and therapeutic approaches worldwide. Since the pivotal clinical trials in 2006 confirmed the therapeutic effect of intravitreal injections of anti-vascular endothelial growth factor (VEGF) antibodies, patients receiving these treatments have experienced marked improvements in vision [15]. However, despite advancements in treatment methods, the overall number of DALYs for low vision and blindness due to AMD continues to increase, indicating that a large number of patients have not fully benefited from existing treatments, which may be related to the aging population and the increasing prevalence of low vision and blindness due to AMD.

Age-stratified analysis reveals that the prevalent cases and DALYs of low vision and blindness due to AMD have substantially increased in the elderly population, especially among those aged 85 and above, where the burden of low vision and blindness due to AMD is the most pronounced. Aging is the key risk factor for AMD, with the macular region of the retina undergoing a series of degenerative changes in the elderly [8]. Studies have indicated that after the age of 60, the incidence of advanced AMD doubles every decade, which can result in severe vision loss and blindness [16]. Further data show that the prevalence of late-stage AMD or ARM in individuals over 75 years old rises significantly to 7.1%, while this rate is markedly lower in younger age groups [17]. These studies consistently emphasize the sharp increase in the incidence and prevalence of advanced AMD with age, which particularly affects the older population and is a leading cause of severe vision loss and blindness [18]. Therefore, regular eye examinations and AMD screenings for the elderly are particularly important, aiding in early detection and intervention, and reducing the risk of vision loss due to AMD. With the intensification of the global aging trend and the continuous rise in the proportion of the elderly population, it is anticipated that the number of patients and disease burden will further increase in the future [19].

Gender analysis indicates that the burden of low vision and blindness due to AMD in females is slightly higher than in males, both in terms of prevalence and DALYs. This finding is consistent with the results of other studies on gender differences in eye diseases and may be caused by a variety of factors [20]. Females generally have a longer life expectancy than males, making them more susceptible to age-related diseases. In some previous studies, females were considered a weak risk factor for late-stage AMD [18]. Additionally, females face more barriers in accessing eye care services, especially in situations with lower levels of education and socioeconomic status [21]. A contributing factor to the temporary increase in ASPR observed in 2020, especially among females, could be the impact of the COVID-19 pandemic, which had a significant effect on healthcare systems worldwide. The pandemic limited access to routine health check-ups and eye care services, potentially delaying the diagnosis and treatment of AMD, and thereby compounding existing challenges faced by females in accessing adequate eye care [22].

Analysis by SDI reveals notable differences in the burden of low vision and blindness due to AMD across regions with varying levels of development. In high SDI regions, despite an increase in the total number of prevalent cases and DALYs, both ASDALYR and ASPR are trending downward, a pattern also demonstrated by other studies[10; 11]. High SDI regions typically have advanced diagnostic and treatment technologies, enabling early screening and timely treatment of AMD, which in turn helps prevent severe outcomes like blindness or significant vision impairment, thereby reducing the disease burden. Studies point out that people in developed countries have easier access to eye care services. The average number of ophthalmologists per million people varies with economic development, from 3.7/million in low-income countries to 76.2/million in high-income countries [23]. Moreover, the spread of health education raises public awareness of AMD and participation in preventive measures, effectively reducing prevalence. Government policy support contrasts with the continued high burden of low vision and blindness due to AMD in low SDI regions, where there is often a lack of necessary diagnostic equipment and professional medical staff, resulting in AMD patients not receiving timely and effective treatment. In low-income and low-SDI regions, ophthalmic care is often overshadowed by other health priorities like infectious disease control, maternal health, and management of chronic conditions such as diabetes [24]. Governments and public health programs tend to prioritize life-threatening illnesses over eye diseases because vision impairment, while disabling, is not typically fatal, thus reducing its perceived urgency in policy decisions [25]. The economic and social impacts of untreated eye conditions are substantial, as they lead to severe vision loss, reducing work capacity and increasing dependence on family members and caregivers, ultimately placing a long-term burden on families and healthcare systems. However, early screening and affordable treatments like bevacizumab (Avastin) can significantly reduce blindness caused by age-related macular degeneration and lower the cost of long-term care [26]. Financial pressure is also an important factor, as many residents cannot afford the high costs of medical treatment, further exacerbating the disease burden. Studies also show that lower levels of education and income are associated with increased incidence and mortality rates of many diseases [27]. For example, individuals with higher levels of education have a slightly reduced risk of developing neovascular AMD, a severe condition that can lead to rapid central vision loss and substantial impairment in daily activities if left untreated [28]. Low income is associated with an increased risk of various diseases that can harm vision; at the same time, poverty may adversely affect professional diagnosis and treatment methods of ophthalmologists [29]. Our study also reports that the ASDALYR and ASPR in the World Bank High Income region are lower than those in the World Bank Low Income, World Bank Lower Middle Income, and World Bank Upper Middle Income regions, further confirming the correlation between the level of economic development and the burden of low vision and blindness due to AMD disease. It is worth noting that the observed increase in low vision and blindness due to AMD prevalence may, at least in part, be attributed to improvements in screening and detection, rather than representing a true rise in prevalence. Screening plays a pivotal role in disease management, with the World Health Organization highlighting its potential to reduce disease burden. However, in low- and middle-income countries, the situation is more complex [30]. While screening enhances disease detection, it does not always translate into effective treatment due to limited healthcare infrastructure and insufficient resources. Regions with better healthcare access tend to implement comprehensive screening programs, which can result in higher reported prevalence. This underscores the importance of considering screening rates when interpreting prevalence data, as improved detection might inflate the apparent burden of low vision and blindness due to AMD. Conversely, in areas with limited healthcare resources and scarce treatment options, the incentive to conduct widespread screening is reduced. As a result, these regions may report lower prevalence rates, not necessarily due to a lower prevalence. Of low vision and blindness due to AMD, but because of underdiagnosis caused by insufficient screening efforts. The lack of uniform screening practices significantly affects the reliability of prevalence estimates, potentially leading to underestimation of disease burden in regions with poor healthcare infrastructure.

There are indeed substantial differences in the burden of low vision and blindness due to AMD across different regions, which may be caused by a variety of factors. Our study found that Asia has the highest number of prevalent cases and DALYs. Asia accounts for over 60% of the world’s population, and thus it is expected to have the highest number of low vision and blindness due to AMD cases. Some studies predict that by 2040, the number of global cases of age-related macular degeneration at any age will rise to 288 million, with the most cases in Asia (reaching 113 million) [5]. The ASDALYR is highest in North Africa and the Middle East. In these regions, AMD accounts for 8.3% of all causes of blindness [31]. The high age-standardized rate in Africa may be partly attributed to the fact that more than two-thirds of affected patients in these regions do not have access to the expensive anti-VEGF treatments that are commonly used in North America and Europe. This economic disparity and the limitation of medical resources result in limited accessibility to AMD treatments, affecting the disease management and vision protection of patients in these regions [32]. In addition, the low vision and blindness due to AMD burden in different latitude regions is comparable and may be related to environmental factors, especially ultraviolet (UV) exposure. Studies have shown that high sunlight exposure may cause damage to the retinal pigment epithelial cells (RPE), which are a crucial part of AMD development. Damage to RPE cells may accelerate degenerative changes in the macular region, thereby increasing the risk of AMD and potentially leading to blindness [12].

In the Sub-Saharan African region, the issue of low vision and blindness caused by AMD is escalating rapidly. This trend is partly due to the fact that approximately 30% of the global population in the Multidimensional Poverty Index resides in Sub-Saharan Africa, where the level of infrastructure investment is among the lowest globally, making eye health issues particularly pronounced. Nonetheless, there have been positive shifts in the region, such as reductions in poverty and increases in life expectancy. The decrease in poverty could positively impact health and alter the disease spectrum in Africa, including the incidence of vision impairment and blindness. However, the anticipated increase in survival rates is likely to be accompanied by an aging population, which will significantly affect the prevalence of age-related eye diseases [33]. Niger and Benin, being part of Sub-Saharan Africa, have the fastest-growing age-standardized rates, attributed to the aforementioned factors. In contrast, the disease burden in South Asia is declining swiftly, with studies indicating that the proportion of blindness caused by macular diseases, including AMD, varies by region, with South Asia having a relatively low proportion (less than 3%) [34].

Differences in the prevalence of AMD among different ethnicities do exist, with higher prevalence in certain groups leading to a correspondingly increased risk of vision loss and blindness.

[35]. Asian populations, particularly the Chinese, have a higher susceptibility to specific subtypes of AMD [36]. Specifically, compared to Caucasians, Asians are more inclined to develop exudative or neovascular AMD [37]. With the continuous growth of China’s elderly population, the number of prevalent cases of low vision and blindness due to AMD has also increased markedly. From 1990 to 2021, the number of patients affected by low vision and blindness due to AMD grew from 882,596 to 2,647,261, making China the country with the highest number of such cases in the world. Studies have indicated that macular degeneration is the third most common cause of moderate to severe vision impairment and the fourth most common cause of blindness in the Chinese population [38]. A study in the Bhaktapur region of Nepal among individuals over 60 years old showed that nearly one-third had some form of AMD. The higher incidence may be significantly associated with smoking habits and a history of cataract surgery [39]. Nonetheless, awareness of AMD in this region is low, with only 7.6% of respondents being knowledgeable about the disease [40]. In a nine-year ophthalmic study conducted in Bhaktapur, the incidence of early AMD was 12.6%, while that of late AMD was 0.7% [41]. Iran has the highest ASDALYR for AMD, indicating a severe burden of the disease in the region. Studies predict that by 2050, the number of AMD cases in Iran may approach 5.5 million, a forecast closely related to Iran being the most populous country in the Middle East and its rapid trend of population aging, which will likely result in a significant increase in vision loss and blindness due to AMD [19].

Technological advancements are transforming the landscape of AMD care worldwide, presenting both opportunities and challenges. In high-income countries (HICs), anti-VEGF therapies such as ranibizumab (Lucentis) and aflibercept (Eylea) are fundamental to AMD treatment, being widely accessible through national healthcare systems and insurance [42]. This accessibility allows for consistent treatment and follow-up, significantly improving patient outcomes and contributing to a reduction in AMD-related blindness in places like the U.K [43].In contrast, there is a notable disparity between HICs and low- and middle-income countries (LMICs). In LMICs, access to these therapies is often limited by high costs, inadequate insurance coverage, and underdeveloped healthcare infrastructure. Even when anti-VEGF drugs are available, their prohibitive prices render them inaccessible to many patients, leading to delays or incomplete treatments that worsen vision loss. Countries like India have explored off-label use of more affordable alternatives, such as bevacizumab (Avastin), but challenges related to treatment frequency, monitoring, and follow-up care persist, thereby diminishing the effectiveness of these alternatives [44]. In a survey conducted in India among patients receiving anti-VEGF treatment, 41% of patients opted for Luteinizing Factor Underlying due to affordability issues. Although this is lower than the 56.7% for ranibizumab, 43.3% of patients still could not afford the lower-cost option, bevacizumab [45]. The uneven implementation of AI diagnostics, telemedicine, and biosimilars further exacerbates the gap between HICs and LMICs. While HICs leverage long-acting anti-VEGF therapies and AI-based screening tools, these innovations are frequently unattainable in resource-constrained environments. Although telemedicine shows potential for enhancing follow-up care in LMICs, issues such as limited internet access and technological literacy pose significant obstacles. To address these disparities, tailored interventions are essential. HICs can focus on adopting emerging therapies like Pegcetacoplan and gene therapies to optimize care, while LMICs would benefit from affordable treatment options, biosimilars, community-based screening programs, and telemedicine [6]. Collaborative partnerships among governments, NGOs, and pharmaceutical companies will be crucial to ensure equitable distribution of technological advancements in AMD care, ultimately preventing further disparities in treatment outcomes.

This study forecasts the ASDALYR and ASPR for low vision and blindness due to AMD from 2022 to 2050. Although the forecast indicates a slight decrease in the ASDALYR and ASPR, the number of prevalent cases and DALYs is expected to continue to increase due to the aging population, a trend consistent with previous studies. For example, one study estimates that by 2050, the number of new cases of early AMD will reach 39.05 million, with 6.41 million new cases of late AMD [13]. Although our study focuses on low vision and blindness due to AMD rather than AMD incidence alone, these projections support our finding that the increasing number of late-stage AMD cases indicates a rise in the number of individuals affected by low vision and blindness due to AMD. Similarly, another study predicts that the global number of people with AMD will rise from 196 million in 2020 to 288 million by 2040 [5]. This increase may be attributed to the intensification of the global aging population, unhealthy lifestyle factors such as smoking and an unbalanced diet, genetic factors, long-term environmental exposure such as ultraviolet radiation, urbanization, and unequal distribution of medical resources, among other factors [9]. Moreover, individuals of lower socioeconomic status may face a higher risk of AMD due to insufficient health awareness, limited access to health services, or inadequate health insurance coverage. Notably, the burden is expected to be higher among females, with both the number of prevalent cases and DALYs projected to exceed those of males throughout the forecast period. The prevalence of visual impairment is consistently higher among females than males, particularly in LMICs. This disparity is driven by a combination of biological and social factors. Biologically, women tend to live longer than men and are more susceptible to AMD. Socially, even when there is no significant difference in disease incidence between genders, societal discrimination may result in higher rates of vision loss among women compared to men [46]. Despite technological advancements in the diagnosis and treatment of AMD, its application may vary by region and may not reach all patients in need. Lack of education and awareness can lead to delayed treatment for patients, affecting the prognosis of the disease. The absence of effective early diagnosis and screening procedures may hinder timely intervention for AMD. Therefore, to more accurately predict the disease burden of low vision and blindness due to AMD and provide policymakers with a basis for developing effective intervention measures, it is necessary to deeply analyze these trends and factors. This will lead to a better understanding of the future trajectory of low vision and blindness due to AMD disease burden and prepare for the challenges ahead.

Strengths and limitations

This study uses ARIMA and ES models to predict the future trends of low vision and blindness due to AMD, offering substantial predictive advantages. The ARIMA model captures complex autoregressive and moving average components, and it often employs differencing to handle non-stationary data effectively. ES models, specifically the damped trend variant of the Holt method, forecast long-term trends by smoothing historical data and assigning higher weights to more recent data points. Our approach not only addresses the complexities of past behaviors through ARIMA but also filters out noise and highlights underlying trends using ES. The combined use of these models enhances the accuracy of predicting the future incidence and burden of low vision and blindness due to AMD, aiding in the formulation of more effective public health strategies and resource allocation plans to better address future health challenges. Moreover, using the 2021 GBD data as a validation set strengthens the reliability and precision of our models. This enables an objective comparison between actual outcomes and model predictions, offering insight into how accurately the models reflect real-world trends.

However, this study has several limitations. First, data quality and completeness are essential for model accuracy; any biases or gaps in data collection may affect the reliability of forecasts. The study’s reliance on the 2021 GBD methodology implies that the absence or inaccuracy of data could introduce biases. The GBD data spans from 1990 to 2021, and this limited time range constrains the foundational data for our models, as longer time spans typically yield higher prediction accuracy. To address forecast uncertainty, we applied a 95% confidence interval to each prediction, providing a range that reflects potential variability based on historical patterns. However, this reliance on past trends assumes continuity, which may limit adaptability in rapidly changing conditions.

Second, while ARIMA and exponential smoothing models manage the complexities of time series, they may not capture all potential influencing factors, such as socioeconomic changes, advancements in medical technology, or variations in patient behavior. Furthermore, the predictive power of these models can be influenced by external events, including global pandemics or other large-scale health crises. To maintain accuracy and reliability, the models’ forecasts should be regularly updated to reflect the latest data and situations. Over time, new information and trends may emerge, necessitating adjustments or recalibrations.

Third, the analysis is based on global aggregate data rather than country-specific trends, potentially limiting the ability to observe variations in disease burden across nations. Although using a single data set streamlines global trend analysis, it may overlook nuances essential to tailored national responses.

Finally, a significant limitation in vision-related research, including this study, is the lack of consideration for the binocular nature of vision. Vision is not merely the sum of each eye’s function; the impact of bilateral eye disease is greater than double that of unilateral disease. This distinction is crucial for interpreting data on visual impairment, particularly in patient-reported outcomes like Health-Related Quality of Life and DALYs. The study does not fully account for this differential impact, which could influence the reported burden of low vision and blindness due to AMD. Future research should incorporate more nuanced measures that reflect the binocular nature of vision to provide a more accurate assessment of the true impact of visual impairment.

Conclusion

In summary, although the overall disease burden of global low vision and blindness due to AMD is increasing, the standardized rates across age groups have declined, indicating progress in addressing this condition worldwide. However, disparities still exist between different genders, sociodemographic indices, and geographical regions. In high SDI regions, despite an increase in the total number of cases, the prevalence and standardized DALY rates are trending downward, suggesting more advanced diagnostic and treatment methods in these areas. In contrast, the burden of low vision and blindness due to AMD remains high in low SDI regions, reflecting a lack of medical resources and public health measures. Particularly among the elderly and women, the burden of low vision and blindness due to AMD is more severe. Going forward, it is necessary to strengthen interventions and preventive measures in these areas and groups with pronounced disparities to further reduce the global disease burden of low vision and blindness due to AMD. Specific measures include increasing resource allocation for eye care, promoting health education programs, and providing financial support to ensure all patients receive the necessary treatment. By continuously monitoring and updating predictive models, policymakers can be provided with more accurate decision-making information, leading to the formulation of more effective public health strategies, ultimately reducing the global burden of low vision and blindness due to AMD and improving the quality of life for the elderly population.

Electronic Supplementary Material

Below is the link to the electronic supplementary material.

Abbreviations

AMD

Age-related Macular Degeneration

GBD

Global Burden of Disease

SDI

Sociodemographic Index

DALY

Annual Disability-Adjusted Life Years

ASPR

Age-Standardized Prevalence Rates

ASDALYR

Age-Standardized DALY Rates

EAPC

Stimated Annual Percentage Change

ARIMA

Autoregressive Integrated Moving Average

UI

Uncertainty Interval

CI

Confidence Interval

LMICs

Low- and Middle-Income Countries

HICs

High-Income Countries

RMSE

Root Mean Square Error

MAPE

Mean Absolute Percentage Error

Author contributions

SYZ and JPR contributed to conception and design of the study. SYZ and SY organized the database. SYZand RTC performed the statistical analysis and wrote the first draft of the manuscript. YS and YZH wrote sections of the manuscript. All authors contributed to manuscript revision, reading, and approved the submitted version.

Funding

This study was funded by Zhejiang Province Traditional Chinese Medicine Science and Technology Program Project (2023ZF034).

Data availability

The data can be accessed and downloaded through the official website of the Institute for Health Metrics and Evaluation (IHME) at http://ghdx.healthdata.org. Given the open-access nature of this database and the absence of personally identifiable information, our study is in compliance with the ethical standards for the use of public data.

Declarations

Ethics approval and consent to participate

This study leveraged publicly available, anonymized data from the Global Burden of Disease Study (GBD 2021), ensuring participant confidentiality and anonymity. As the data is open-access and devoid of personally identifiable information, our research was exempt from full ethical review.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.