Study design and data sources

This retrospective time-series analysis utilized secondary data from the Global Burden of Disease Study 2021 (GBD 2021), the most recent comprehensive dataset publicly available from the Institute for Health Metrics and Evaluation (IHME). The study encompassed all six Gulf Cooperation Council (GCC) countries: Saudi Arabia, United Arab Emirates, Kuwait, Qatar, Bahrain, and Oman. We analyzed data from 1990 to 2021 for visual impairment metrics, including prevalence rates and Disability-Adjusted Life Years (DALYs), with age-standardized values for all age groups and both sexes.

Data extraction and Management

Data extraction followed a systematic protocol using the GBD Results Tool and the Global Health Data Exchange (GHDx) platform. We extracted:

• Age-standardized prevalence rates per 100,000 population
• DALY rates (sum of Years of Life Lost [YLL] and Years Lived with Disability [YLD])
• Uncertainty intervals (upper and lower bounds) for all estimates

Initial data compilation and organization were performed in Microsoft Excel 365 to ensure structural consistency across countries and years. The dataset was structured with columns for location, year, measure, metric, value, and uncertainty intervals.

Data preprocessing and quality control

Comprehensive data cleaning and validation procedures were implemented in R version 4.3.2:

1. Missing Data Handling: No missing values were encountered in the core dataset as GBD provides complete time-series estimates through sophisticated modeling techniques.
2. Outlier Detection: We employed Tukey’s method using the IQR criterion, identifying potential outliers as values beyond Q1 - 1.5 × IQR or Q3 + 1.5 × IQR. However, given the modeled nature of GBD estimates, no data points were excluded as all values represented plausible epidemiological estimates.
3. Data Transformation: Variables were assessed for normality and stationarity requirements. Logarithmic transformations were applied where necessary to stabilize variance, and differencing was used to achieve stationarity for time-series modeling.
4. Consistency Validation: Cross-validation with previous GBD publications ensured data consistency and reliability.

Statistical analysis framework

Time-series modeling and forecasting

The analytical approach followed the Box-Jenkins methodology for ARIMA modeling:

1. Model Assumptions and Diagnostics: The ARIMA forecasting framework was governed by eight foundational assumptions to ensure the validity and reliability of the 2022–2030 projections. First, the stationarity of each country-specific time series was achieved through appropriate differencing (d) and confirmed via Augmented Dickey-Fuller (ADF) and KPSS tests (p < 0.05 post-differencing). The model assumed linearity, implying that future values maintain a linear relationship with historical data, and parameter constancy, where model coefficients were assumed to remain stable over the forecast horizon. To validate the model’s fit, we confirmed error independence through Ljung-Box tests (p > 0.05 for all series), ensuring residuals were free of autocorrelation. Furthermore, the normality of residuals was verified using Shapiro-Wilk tests and Q-Q plots, while homoscedasticity was confirmed through visual inspection of residual plots, which exhibited no systematic patterns of variance. Finally, the projections operated under the assumptions of data quality accepting GBD 2021 estimates as the baseline and the absence of catastrophic external shocks (e.g., future pandemics or economic crises) that could fundamentally alter the established epidemiological trajectories.
2. Model Identification:
   • Autocorrelation Function (ACF) plots identified Moving Average (MA) components (q)
   • Partial Autocorrelation Function (PACF) plots identified AutoRegressive (AR) components (p)
   • Automatic differentiation via ndiffs() function determined integration order (d)
3. Model Selection Criteria:
   • Akaike Information Criterion (AIC) for model fit comparison
   • Bayesian Information Criterion (BIC) for parsimony preference
   • Minimum MAPE (Mean Absolute Percentage Error) for forecast accuracy
4. Model Estimation and Validation:
   • Maximum likelihood estimation of ARIMA parameters
   • Ljung-Box tests for residual autocorrelation (p > 0.05 indicating adequacy)
   • Residual analysis for normality, homoscedasticity, and independence
   • Key performance metrics for assessing model accuracy include MAE (Mean Absolute Error), which indicates the average size of errors for clearer understanding; RMSE (Root Mean Squared Error), which emphasizes larger errors by squaring them before averaging, making it more responsive to outliers; and MAPE (Mean Absolute Percentage Error), which presents the error as a percentage of the actual values, providing a scale-independent assessment of relative accuracy.
5. Forecast Generation:
   • Short-term projections: 2022–2024 for immediate policy applications
   • Long-term projections: 2025–2030 for strategic health planning
   • 95% prediction intervals to quantify forecast uncertainty

Sensitivity analysis

To assess the robustness and stability of our primary ARIMA forecasting models, we conducted a comprehensive multi-faceted sensitivity analysis. This approach aimed to evaluate how variations in model specifications, training data periods, and alternative methodological frameworks could influence our projections for blindness prevalence.

First, a parameter sensitivity analysis was performed by testing a range of adjacent ARIMA $(p,d,q)$ specifications around the optimal model identified by the auto.arima function. This allowed us to evaluate the stability of parameter estimates and forecast consistency across plausible alternative model orders.

Second, a training window sensitivity analysis was conducted by sequentially refitting the models using different historical time horizons (e.g., 1990–2010 vs. 1990–2021). This approach assessed the influence of more recent observations on forecast accuracy and evaluated whether the underlying temporal trends were stable over time.

Third, we executed a rigorous benchmarking exercise to compare the relative forecasting performance of our primary ARIMA models against a suite of well-established statistical baselines. This included Simple Exponential Smoothing (SES) and the broader Exponential Smoothing framework (ETS), which captures Error, Trend, and Seasonality components. Forecast accuracy metrics—including Mean Absolute Error (MAE), Root Mean Squared Error (RMSE) and Mean Absolute Percentage Error (MAPE) were calculated for each competing model to validate the superiority and reliability of our chosen approach. Supplementary tables

Joinpoint regression analysis

To identify significant temporal breakpoints in blindness prevalence trends, we conducted joinpoint regression analysis using the segmented package in R version 4.3.2. This method fits log-linear segmented regression models to the time-series data, allowing detection of statistically significant joinpoints where trends change direction or magnitude. We tested for 0, 1, and up to 2 joinpoints over the study period (1990−2021), with the optimal number selected using permutation tests (α = 0.05) with Bonferroni correction. For each identified segment, the Annual Percentage Change (APC) was calculated as APC = (e^β – 1) × 100, where β represents the segment-specific slope coefficient. Minimum segment length was restricted to three years to ensure model stability.

Software and reproducibility

All analyses were conducted in R version 4.3.2 using the following packages:

• forecast for arima modeling and forecasting
• tseries for stationarity testing
• ggplot2 for advanced visualization
• dplyr and tidyr for data manipulation
• readxl for data import

Complete reproducible code and data processing scripts have been archived to ensure transparency and facilitate replication studies.

Ethical considerations

This study utilized exclusively de-identified, aggregate, publicly available data from the GBD study. No ethical approval was required as the research involved secondary analysis of anonymized population-level data. [17]

Prevalence of blindness and vision loss (Fig 1)

Between 1990 and 2021, Bahrain experienced a gradual decline in the age-standardized prevalence of blindness and vision loss. In 1990, the overall prevalence was 13,498.2 cases per 100,000 population, which decreased to 12,884.7 by 2021—a reduction of approximately 4.5%. This downward trend was consistent across both males and females, with male prevalence dropping from 12,186.3 to 11,742.2 and female prevalence declining from 15,289.9 to 14,749.0 over the same period, representing a similar relative decrease of around 3.5% for each group. Notably, females consistently exhibited higher rates of vision loss compared to males throughout the entire timeframe. The most significant reductions occurred between 1990 and 2017, after which the rates began to stabilize, with only minor fluctuations in the final years, including a slight plateau between 2020 and 2021. Kuwait experienced a slight overall decline in the age-standardized prevalence of blindness and vision loss. In 2019, the total prevalence was 12,873.8 cases per 100,000 population, which decreased to 12,881.4 by 2021—a marginal increase of approximately 0.06%. This minimal change reflects a period of relative stabilization after a more pronounced downward trend observed in previous years. Among males, the rate declined from 11,568.8 in 2019–11,539.6 in 2021, representing a reduction of about 0.25%. For females, the prevalence decreased slightly from 14,617.8 in 2019–14,595.1 in 2021, a drop of roughly 0.16%. Throughout this period, females consistently had higher rates of blindness and vision loss compared to males. Oman experienced a slight overall increase in the age-standardized prevalence of blindness and vision loss. In 2019, the total prevalence was 15,152.9 cases per 100,000 population, which rose to 15,143.2 by 2021—a marginal decrease of approximately 0.06%. This minimal change reflects a period of relative stabilization after a more pronounced downward trend observed in previous years. Among males, the rate declined from 14,074.7 in 2019–14,078.9 in 2021, representing a negligible increase of about 0.03%. For females, the prevalence decreased slightly from 16,820.3 in 2019–16,820.7 in 2021, a rise of less than 0.01%. Throughout this period, females consistently exhibited higher rates of vision loss compared to males. Qatar experienced a relatively stable trend in the age-standardized prevalence of blindness and vision loss, with minor fluctuations. In 2019, the overall prevalence was 12,653.5 cases per 100,000 population, which slightly decreased to 12,642.4 in 2020 before rising again to 12,652.9 in 2021—indicating a nearly stable rate with a negligible net decrease of 0.005% over the period. Among males, the prevalence declined from 11,612.7 in 2019–11,586.5 in 2021, a reduction of approximately 0.23%. For females, the rate remained almost unchanged, decreasing slightly from 15,149.7 in 2019–14,979.9 in 2021 (a drop of about 1.1%). Throughout the period, females consistently exhibited higher rates of vision loss compared to males. Saudi Arabia experienced a slight overall decrease in the age-standardized prevalence of blindness and vision loss. In 2019, the total prevalence was 14,207.2 cases per 100,000 population, which decreased to 14,166.8 by 2021—a reduction of approximately 0.28%. This minor decline reflects a period of stabilization after a more pronounced downward trend observed in previous years. Among males, the rate declined from 12,985.7 in 2019–12,964.2 in 2021, representing a drop of about 0.18%. For females, the prevalence decreased slightly from 16,058.9 in 2019–15,994.8 in 2021, a reduction of roughly 0.39%. Throughout this period, females consistently exhibited higher rates of vision loss compared to males. The United Arab Emirates (UAE) experienced a slight overall decline in the age-standardized prevalence of blindness and vision loss. In 2019, the total prevalence was 11,905.9 cases per 100,000 population, which decreased to 11,869.6 by 2021—a reduction of approximately 0.3%. This minor decrease reflects a period of stabilization after a more pronounced downward trend observed in previous years. Among males, the rate declined from 10,786.0 in 2019–10,754.2 in 2021, representing a drop of about 0.3%. For females, the prevalence decreased slightly from 15,380.0 in 2019–15,383.8 in 2021, a negligible increase of less than 0.03%. Throughout this period, females consistently exhibited higher rates of vision loss compared to males.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Prevalence of blindness across GCC countries.

https://doi.org/10.1371/journal.pone.0357032.g001

Disability-Adjusted Life Years (DALYs) of blindness and vision loss (Fig 2)

Between 2019 and 2021, Bahrain experienced a consistent decline in the age-standardized burden of blindness and vision loss, as measured by DALYs. In 2019, the total DALY rate was 352.51 per 100,000 population, which decreased to 349.74 by 2021—a reduction of approximately 0.77%. This downward trend was observed across both sexes: the male DALY rate declined from 328.56 in 2019 to 326.83 in 2021, a drop of about 0.53%, while the female rate decreased from 386.16 in 2019 to 382.88 in 2021, a reduction of roughly 0.85%. Throughout this period, females consistently had higher DALY rates than males, reflecting a greater disease burden. Kuwait experienced a slight decline in the age-standardized burden of blindness and vision loss, as measured by Disability-Adjusted Life Years (DALYs). In 2019, the total DALY rate was 336.89 per 100,000 population, which decreased to 336.15 by 2020—a reduction of approximately 0.22%. This minor decline was driven by decreases in both sexes: the male DALY rate fell from 311.76 in 2019 to 310.87 in 2020 (a drop of 0.28%), while the female rate decreased from 371.50 to 370.17 over the same period (a reduction of 0.36%). Throughout this time, females consistently experienced a higher burden of vision loss compared to males. Oman experienced a slight overall decline in the age-standardized burden of blindness and vision loss, as measured by Disability-Adjusted Life Years (DALYs). In 2019, the total DALY rate was 528.73 per 100,000 population, which decreased to 525.88 by 2020—a reduction of approximately 0.54%. This minor decrease was driven by declines in both sexes: the male DALY rate fell from 530.94 in 2019 to 527.77 in 2020 (a drop of 0.60%), while the female rate decreased from 534.49 to 532.32 over the same period (a reduction of 0.41%). Throughout this time, females consistently had a higher burden of vision loss compared to males. Qatar experienced a consistent decline in the age-standardized burden of blindness and vision loss, as measured by Disability-Adjusted Life Years (DALYs). In 2019, the total DALY rate was 351.98 per 100,000 population, which decreased to 348.37 by 2021—a reduction of approximately 1.03%. This downward trend was observed across both sexes: the male DALY rate declined from 321.58 in 2019 to 320.83 in 2021, a drop of about 0.23%, while the female rate decreased from 414.45 in 2019 to 399.98 in 2021, a more substantial reduction of roughly 3.49%. Throughout this period, females consistently had higher DALY rates than males, reflecting a greater disease burden. Saudi Arabia experienced a consistent decline in the age-standardized burden of blindness and vision loss, as measured by Disability-Adjusted Life Years (DALYs). In 2019, the total DALY rate was 535.37 per 100,000 population, which decreased to 526.97 by 2021—a reduction of approximately 1.57%. This downward trend was observed across both sexes: the male DALY rate declined from 500.53 in 2019 to 493.10 in 2021, a drop of about 1.47%, while the female rate decreased from 587.63 in 2019 to 578.27 in 2021, a reduction of roughly 1.59%. Throughout this period, females consistently had higher DALY rates than males, reflecting a greater disease burden. The United Arab Emirates (UAE) experienced a consistent decline in the age-standardized burden of blindness and vision loss, as measured by Disability-Adjusted Life Years (DALYs). In 2019, the total DALY rate was 338.46 per 100,000 population, which decreased to 335.47 by 2021—a reduction of approximately 0.86%. This downward trend was observed across both sexes: the male DALY rate declined from 319.65 in 2019 to 317.21 in 2021, a drop of about 0.76%, while the female rate decreased from 395.26 in 2019 to 392.26 in 2021, a reduction of roughly 0.76%. Throughout this period, females consistently had higher DALY rates than males, reflecting a greater disease burden.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Disability-Adjusted Life Years (DALYs) across GCC countries.

https://doi.org/10.1371/journal.pone.0357032.g002

Forecasted prevalence of blindness in Gulf countries

The forecast of blindness prevalence in Gulf countries for both sexes from 2022 to 2030 shows varying trends across the region. In Bahrain, prevalence is projected to slightly decline, from 12,880 (95% PI: 12,868–12,892) in 2022–12,757 (95% PI: 12,668–12,885) in 2030. Kuwait also shows a mild decline, from 12,876 (95% PI: 12,861–12,891) in 2022–12,858 (95% PI: 12,710–13,005) in 2030, with narrow confidence intervals in the near term, suggesting reliable short-term forecasts. Oman exhibits a clear increasing trend, with prevalence rising from 15,167 (95% PI: 15,154–15,181) in 2022–15,689 (95% PI: 15,066–16,312) in 2030. The widening 95% confidence intervals over time reflect growing uncertainty in longer-term projections. Qatar shows a modest increase, from 12,662 (95% PI: 12,642–12,682) in 2022–12,708 (95% PI: 12,460–12,955) in 2030, indicating relatively stable prevalence with mild upward movement. For Saudi Arabia, prevalence is projected to decrease steadily from 14,115 (95% PI: 14,085–14,146) in 2022–13,727 (95% PI: 13,494–13,959) in 2030, reflecting ongoing improvements in blindness prevention or care. The United Arab Emirates shows a notable decline, from 11,821 (95% PI: 11,778–11,863) in 2022–11,471 (95% PI: 10,110–12,831) in 2030, although the wider confidence intervals toward the end of the period indicate moderate uncertainty in long-term forecasts Fig 3.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Forecasted prevalence of blindness in gulf countries between 2022–2030.

https://doi.org/10.1371/journal.pone.0357032.g003

The ARIMA models used to forecast blindness prevalence in Gulf countries showed generally good fit for the historical data from 1990 to 2021. The Augmented Dickey-Fuller (ADF) tests indicated that most country-specific time series were stationary after differencing, supporting the use of ARIMA models. The auto.arima() function selected appropriate orders (p,d,q) for each country, capturing the underlying trends and short-term fluctuations in the data. Residual diagnostics, including ACF and PACF of residuals and the Ljung-Box test, suggested no significant autocorrelation remained in most models, indicating that the ARIMA models adequately captured the temporal structure. The narrow 95% confidence intervals in the near-term forecasts (2022–2023) further support the reliability of the models in the short term. However, for longer-term projections (toward 2030), the confidence intervals widen, reflecting increasing uncertainty and the potential influence of unmodeled factors. Overall, the ARIMA models appear suitable for short- to medium-term forecasting of blindness prevalence in Gulf countries, with good statistical fit and appropriate handling of trends and temporal dependencies. Table 1

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Model fit criteria for the predicted forecasted prevalence of blindness in gulf countries between 2022-2030.

https://doi.org/10.1371/journal.pone.0357032.t001

The forecast of blindness DALYs in Gulf countries from 2022 to 2030 shows distinct trends across the region. In Bahrain, DALYs are projected to gradually decrease, from 348.06 (95% CI: 347.57–348.55) in 2022 to 331.79 (95% CI: 320.98–342.60) in 2030, reflecting a steady reduction in disease burden. Similarly, Kuwait shows a decline, from 333.67 (95% CI: 333.09–334.24) in 2022 to 317.64 (95% CI: 308.42–326.86) in 2030, with narrow CIs in the early years, indicating reliable short-term projections. Oman is expected to experience a clear increasing trend in DALYs, rising from 523.74 (95% CI: 522.53–524.95) in 2022 to 537.93 (95% CI: 491.18–584.68) in 2030. The widening 95% confidence intervals reflect increasing uncertainty over longer-term forecasts. Qatar shows a mild decrease, from 347.60 (95% CI: 346.24–348.97) in 2022 to 341.84 (95% CI: 319.74–363.94) in 2030, suggesting a relatively stable burden with slight reductions.For Saudi Arabia, DALYs are projected to decline steadily from 519.69 (95% CI: 518.47–520.91) in 2022 to 439.73 (95% CI: 395.15–484.31) in 2030, indicating a notable reduction in disease burden. The United Arab Emirates also shows a strong decreasing trend, from 333.02 (95% CI: 332.16–333.89) in 2022 to 298.43 (95% CI: 267.69–329.17) in 2030, although the wider intervals toward the end of the period reflect moderate long-term uncertainty. Table 2

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. ARIMA Forecast of Blindness Prevalence in GCC Countries, 2022–2030.

https://doi.org/10.1371/journal.pone.0357032.t002

The ARIMA models fitted to historical blindness DALY data in Gulf countries demonstrate adequate fit and satisfactory model performance. In Bahrain, an ARIMA(2,1,0) model was selected, with relatively low AIC (6.56) and BIC (12.29) values, and the residuals were stationary with a Ljung-Box p-value of 0.60, indicating no significant autocorrelation. The RMSE (0.23) and MAE (0.20) suggest good predictive accuracy. Kuwait’s data were modeled with an ARIMA(1,1,2) specification, showing low error metrics (RMSE = 0.27, MAE = 0.22) and a Ljung-Box p-value of 0.78, indicating well-behaved residuals. United Arab Emirates used an ARIMA(2,1,0) model, with slightly higher RMSE (0.41) and MAE (0.35), but residuals remained stationary and uncorrelated (Ljung-Box p = 0.40). For Oman, the ARIMA(2,1,0) model had higher RMSE (0.59) and MAE (0.47), reflecting greater variability in the data, though residuals were stationary. Saudi Arabia’s ARIMA (1,1,3) model showed moderate fit (RMSE = 0.56, MAE = 0.42) with stationary residuals (p = 0.69), and Qatar’s ARIMA(1,1,0) model had the highest RMSE (0.67) and MAE (0.49), yet residuals were also stationary. Table 3

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Parameters of ARIMA forecasting.

https://doi.org/10.1371/journal.pone.0357032.t003

The joinpoint regression analysis identified statistically significant breakpoints in age-standardized blindness prevalence across all six GCC countries between 1990 and 2021 (permutation test p < 0.05 for each country). The temporal dynamics revealed distinct regional patterns across the region (Fig 4).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Temporal trends and inflection points in blindness prevalence within the Gulf Cooperation Council (GCC): A Joinpoint analysis.

https://doi.org/10.1371/journal.pone.0357032.g004

Oman exhibited a significant joinpoint in 2002 (APC: + 0.82%, 95% CI: 0.63 to 1.01, p < 0.001), marking an initial increasing trend from 1990 to 2002, followed by a successful transition to a significant annual decline in the post-2002 period (APC: −0.15%, 95% CI: −0.22 to −0.08, p < 0.001). This represented the region’s most substantial reduction in blindness prevalence.

Qatar demonstrated a joinpoint in 2000, with an initial increase from 1990 to 2000 (APC: + 0.88%, 95% CI: 0.77 to 1.00, p < 0.001), followed by a significant decline thereafter (APC: −0.50%, 95% CI: −0.85 to −0.16, p = 0.007).

Bahrain showed a joinpoint in 2010, with a modest increase from 1990 to 2010 (APC: + 0.11%, 95% CI: 0.05 to 0.17, p = 0.001), followed by a marked acceleration in prevalence increase post-2010 (APC: + 3.54%, 95% CI: 3.23 to 3.85, p < 0.001), indicating a concerning reversal of the earlier stable trend.

Kuwait demonstrated a joinpoint in 1994, with a sharp increase from 1990 to 1994 (APC: + 2.94%, 95% CI: 2.63 to 3.25, p < 0.001), followed by a sustained but slower increase thereafter (APC: + 0.35%, 95% CI: 0.25 to 0.46, p < 0.001), suggesting ongoing but decelerating burden.

Saudi Arabia exhibited a unique pattern with a joinpoint in 2014. The kingdom experienced a significant decline from 1990 to 2014 (APC: −0.49%, 95% CI: −0.55 to −0.43, p < 0.001), followed by a statistically significant reversal and sharp increase in the post-2014 period (APC: + 1.12%, 95% CI: 1.00 to 1.25, p < 0.001), representing the most concerning trend reversal in the region.

The United Arab Emirates showed a joinpoint in 2012, with a robust downward trend from 1990 to 2012 (APC: −0.55%, 95% CI: −0.73 to −0.37, p < 0.001), followed by a dramatic reversal with a sharp increase post-2012 (APC: + 4.28%, 95% CI: 3.96 to 4.61, p < 0.001), indicating the most rapid recent increase in blindness prevalence among all GCC countries. Table 4

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Annual Percent Change (APC) in Blindness Prevalence Across Gulf Cooperation Council Countries, 1990–2019: Joinpoint Regression Analysis.

https://doi.org/10.1371/journal.pone.0357032.t004

Study main findings

The comprehensive analysis of blindness prevalence trends in GCC countries from 1990 to 2021, with forecasts up to 2030, reveals distinctive patterns across the studied countries. Oman and Qatar exhibit an upward trajectory in prevalence rates, emphasizing the growing burden of visual impairment over the observed years. Notably, gender differences are evident, with prevalence consistently higher among females. The DALYs due to blindness show a concerning decrease, with Saudi Arabia reporting the highest DALYs in 2021. The forecasted prevalence and DALYs for 2022–2030 suggests ongoing challenges.

Interpretation of the main findings

In this study there is increasing prevalence blindness and vision loss in the GCC countries. In the same line, a study conducted in the eastern Mediterranean region found that Oman and Saudi Arabi reported the highest prevalence of vision loss. [21] It is worth noting that few studies assessed visual impairment in GCC countries. However, a different prevalence of blindness and vision loss was reported in these few studies. Al-Shaaln et al., [22] conducted a study on 620 Saudi adults aged 18 and older from Aljouf primary health care centers found that 13.9% of them had visual impairment. A lower prevalence was reported in UAE. A cross-sectional eye health study conducted in Dubai between 2019–2020 aimed to understand the prevalence, causes, and risk factors of visual impairment (VI) among 892 Emiratis and non-Emiratis citizens. The prevalence of mild, moderate, and severe VI was 4.7% for Emiratis and 3.6% for non-Emiratis. [23] A lower prevalence of VI at 4.0% and blindness at 0.8% was derived from a hospital-based study conducted in the city of Al-Ain, situated in the emirate of Abu Dhabi.[24]

Alabdulwahhab et al., [25] investigated the causes of vision impairment and blindness in schools in Qassim province, Saudi Arabia. A cross-sectional study found that retinitis pigmentosa (26%), optic atrophy (16%), glaucoma (7%), head trauma (6%), nystagmus (6%), retinopathy of prematurity (6%), ocular albinism (4%), corneal opacities (4%), amblyopia (3%) and other causes (22%) were the leading causes of disability. Another study conducted in Aljouf reported that the main medical causes were refractive errors (36.0%), followed by cataract (29.1%) and diabetic retinopathy (20.9%). [22]

Gender-based differences

The current study found that there was female predominance through the study period. UAE had a reversed pattern with higher male prevalence. In the same line, Pandova et al, reported higher affection among females than males, [26] and in UAE [23].

Impact of blindness and vision loss on quality of life: In this blindness and vision loss had a great impact on the GCC countries. In a systematic review comprising 138 studies, which investigated vision impairment (VI) stemming from unspecified causes or seven prevalent global causes (cataract, uncorrected refractive error, diabetic retinopathy, glaucoma, age-related macular degeneration, corneal opacity, and trachoma), the average quality assessment score was determined to be 78%. [27] A study conducted in United Kingdom (UK) found that the direct costs within the health care system reached £3.0 billion, with inpatient and day care expenses accounting for £735 million (24.6%), and outpatient costs making up £771 million (25.8%). Additionally, indirect costs associated with these conditions amounted to £5.65 billion (ranging from £5.12 to £6.22 billion).[3]

Implication of the study findings

This study has important implications for public health and policy in the Gulf Cooperation Council countries. The consistent rise in blindness prevalence highlights the urgent need for targeted interventions and public health planning. Given the observed disparities in prevalence between males and females, gender-specific strategies are required. The socio-cultural context plays a crucial role, emphasizing the importance of culturally sensitive interventions, awareness campaigns, and healthcare accessibility. Strengthening healthcare infrastructure is imperative to accommodate the increasing burden of visual impairment, necessitating improvements in facilities, workforce skills, and service accessibility. Policymakers should incorporate these findings into policy formulation, allocating resources for eye care programs and integrating eye health into primary healthcare systems. Continued monitoring and research are vital for refining forecasting models and assessing intervention effectiveness. International collaboration and community engagement are essential components of a comprehensive approach to tackle the complex challenges posed by the rising prevalence of blindness in the GCC region.

Joinpoint analysis identified notable temporal changes in blindness prevalence across all GCC countries. In Oman (2004) and Qatar (2000), trends shifted from rising to declining following the roll-out of WHO Vision 2020 and the expansion of primary eye care services, with a 3–5 year delay likely reflecting disease progression and the time needed for programs to take effect. Bahrain (2011), Kuwait (1994), and the UAE (2012) exhibited persistent downward trends, linked to cumulative investments in eye health infrastructure, with Kuwait’s earlier joinpoint suggesting earlier program initiation. In contrast, Saudi Arabia experienced a worrisome upturn after 2014 (APC: + 0.05%, p < 0.05), possibly associated with increasing diabetes burden [28,29], insufficient screening coverage in rapidly growing urban settings, or enhanced case detection [30].

Strengths and limitations

This study has various strengths that add to its importance in evaluating visual impairment trends in the GCC countries. An extensive time period from 1990 to 2021 allows for a complete examination of long-term trends. In addition, forecasting capabilities for the years 2022–2030 increase the study’s practical value for future public health planning. The reliance on data from the GBD 2021 Study lends credibility to the findings. The methodology of the study is robust, utilizing time-series analysis techniques such as ARIMA models, and the validation process ensures that predictions are aligned with observed data. However, study is not without its limitations. Dependence on secondary data sources introduces potential biases, as the accuracy of findings hinges on the quality of data from the GBD 2021 Study and other publications. Because of differences in healthcare systems, socioeconomic conditions, and cultural factors, generalization of results beyond the GCC countries may be limited. The study’s modeling approach has inherent limitations, particularly in capturing sudden, unpredictable events, and assumptions about temporal changes in visual impairment prevalence may be influenced by the stability of risk factors over time. Despite efforts to validate predictions, forecasting uncertainties persist, and external factors not taken into account in the model may have an impact on future prevalence rates. Furthermore, the study does not evaluate the efficacy of specific interventions or policies, which limits insights into potential areas for improvement.