Socioeconomic status mediates and modifies racial and ethnic disparities in incisional glaucoma surgical outcomes.

Purpose: To estimate the proportion of racial and ethnic disparities observed in glaucoma surgical outcomes that can be eliminated by curbing differences in socioeconomic status (SES).

Design: Retrospective cohort study.

Subjects: The entire population of 2016-2018 California (CA) fee-for-service Medicare beneficiaries with a claim for incisional glaucoma surgery (trabeculectomy, tube shunt, or EX-PRESS shunt).

Methods: The primary exposure was race and ethnicity, stratified into: Non-Latinx White (as the reference category), Black, Latinx, Asian/Pacific Islander (PI), and Other. The SES mediator was dichotomized to low vs. non-low based on dual-eligibility for Medicaid. Outcome (time to failure event) was defined as having a claim for a glaucoma surgery revision or reoperation event.

Main Outcome Measures: The total effect (TE) estimated the entire racial and ethnic disparity. The controlled direct effect (CDE) estimated the remaining disparity after fixing SES to non-low for all, and the proportion eliminated (PE) estimated the proportion of the disparity eliminated after uniform SES assignment. The TE and CDE estimates are interpreted as hazards ratios given time-to-event modeling using Cox proportional hazards.

Results: The final analytical sample included a total of 5,985 unique CA beneficiaries. After uniformly fixing SES to non-low, the racial and ethnic disparity for Black patients dissipated most (TE: 1.18, 95% CI: 0.99-1.41; CDE: 1.01, 95% CI: 0.80-1.77), followed by Latinx patients (TE: 1.23, 95% CI: 1.08-1.51; CDE: 1.10, 95% CI: 0.90-1.35), Other race and ethnicity patients (TE: 1.32, 95% CI: 1.03-1.70; CDE: 1.24, 95% CI: 0.91-1.68), and Asian/PI patients (TE: 1.18, 95% CI: 1.02-1.36; CDE: 1.21, 95% CI: 0.99-1.47). The PE estimates suggest that equalizing SES would eliminate varying levels of disparities, with a maximum of 96% for Black patients (PE: 0.96, 95% CI: -0.27-2.19), followed by 54% for Latinx patients (PE: 0.54, 95% CI: -0.25-1.33), and no significant change for Other race and ethnicity (PE: 0.24, 95% CI: -0.49-0.97), and Asian/PI patients (PE: -0.18, 95% CI: -1.11-0.75).

Conclusions: We found that SES mediates racial and ethnic disparities in glaucoma surgical outcomes, though by varying amounts by individual racial and ethnic group. Of note, addressing SES differences would eliminate 96% of the disparity for Black beneficiaries.