INTRODUCTION:
Eye health disparities in South Africa are well documented across gender, residence, race, and socioeconomic status. Although previous studies have linked eye health outcomes to multiple factors, few quantify the relative contribution of specific covariates to these disparities. This lack of evidence limits informed resource allocation and the design of targeted interventions to reduce inequities. This study addressed this gap by decomposing the rural-urban divide in eye health outcomes, separating the share attributable to differences in population characteristics from the share driven by differential effects of these characteristics between rural and urban groups.

METHODS:
Using a cross-sectional design, data from SANHANES were analysed using the Oaxaca-Blinder decomposition method to quantify rural-urban disparities in eye health outcomes. The analysis identified the contributions of differences in population characteristics (endowments) and differences in their effects (coefficients) to disparities in eye examination uptake, predictors of vision loss, and the use of distance and near vision corrective glasses.

RESULTS:
Persistent rural-urban disparities in eye health outcomes were evident, though their underlying mechanisms differed by outcome. Province, wealth, race and diabetes diagnosis accounted for much of the gap in eye exam uptake explained by endowments, albeit insignificant overall, while locality, race, age and wealth accounted for the gap explained by coefficients (significant overall). Disparities in vision loss identification, as well as in distance vision eyeglasses usage showed no statistically significant contributions from endowments or coefficients. Wealth, race, and province contributed significantly to disparities in near vison eyeglasses use.

CONCLUSIONS:
No covariate consistently dominated across outcomes, indicating that rural-urban disparities may be driven by other unobserved structural and contextual factors. Future research should explore these determinants using mixed methods approaches, that possibly incorporate ethnography.