The Role of Geography in United States Retinoblastoma Care

PURPOSE: To evaluate the association between geography and retinoblastoma care, including treatment modality, stage at diagnosis, and time to treatment in the United States, and whether geography modifies ethnoracial and socioeconomic disparities in care. DESIGN: Retrospective, population-based cohort study. SUBJECTS: Children aged <15 years diagnosed with retinoblastoma from 2000 to 2021 in the Surveillance, Epidemiology, and End Results (SEER) Program of the National Cancer Institute. METHODS: We examined the association between geography and treatment modality, stage at diagnosis, and time to treatment, while accounting for race/ethnicity and county-level median household income. Multinomial and binary logistic regression models were used to evaluate predictors of treatment modality and stage at diagnosis, and multiple linear regression was used to assess time from diagnosis to treatment initiation (days). Interaction terms between geography and race/ethnicity, as well as between geography and income were included to assess effect modification. MAIN OUTCOME MEASURES: Adjusted odds ratios (aORs) and 95% confidence intervals (CIs) for associations with treatment modality, stage at diagnosis, and time to treatment. RESULTS: The cohort included 1375 pediatric patients (1817 eyes). Children residing in metropolitan areas had lower odds of enucleation (aOR: 0.692; 95% CI: 0.451-0.914; P = .002) and reduced receipt of chemotherapy (aOR: 0.629, 95% CI: 0.444-0.891, P = .015). Non-Hispanic White children also had reduced odds of enucleation compared with non-White children (aOR: 0.757; 95% CI: 0.613-0.838; P = .010). Interaction analyses demonstrated that geography modified associations between race/ethnicity and enucleation (P = .018) and receipt of chemotherapy (P = .031). Among patients with staging data, those from lower-income households were more likely to present with regional/distant disease (aOR: 1.342; 95% CI: 1.010-1.794; P = .044), while Non-Hispanic White children were more likely to present with localized disease (aOR: 1.931; 95% CI: 1.303-2.863; P = .001). Geography modified the association between household income and stage at presentation (P = .022). Time to treatment did not differ significantly across patients, regardless of geography. CONCLUSIONS: Geography influences disparities in retinoblastoma care by shaping access to key therapies and modifying the impact of race/ethnicity and socioeconomic status. With no differences in time to treatment after diagnosis, inequities may arise earlier in the care pathway, during disease recognition and referral. Strengthening early detection, referral networks, and access to advanced therapies in non-metropolitan regions may help reduce these disparities.