Geographic distance and overall survival among patients receiving non-palliative radiotherapy: a diagnosis-stratified real-world cohort study

To determine whether residential distance to a tertiary radiotherapy centre is associated with overall survival among patients receiving non-palliative radiotherapy, while accounting for diagnostic heterogeneity and repeated treatment episodes. We conducted a retrospective single-centre cohort study of patients receiving non-palliative radiotherapy for six prevalent cancer groups. The first eligible treatment episode per patient defined the primary cohort. Orthodromic distance was evaluated in five categories, continuously and with restricted cubic splines. Cox regression was stratified by diagnosis and adjusted for age, sex and mobility. Secondary analyses examined clinical-variable availability, models with available stage and/or ECOG, diagnosis-by-distance heterogeneity, treatment intent, fractionation, treatment year and all eligible episodes with patient-clustered robust standard errors. The primary analysis included 5,690 patients and 1,012 deaths. Median distance was 14.4 km; 94.3% lived within 60 km. There was no evidence of an association between categorical distance and mortality in the diagnosis-stratified model adjusted for available measured covariates (global P =0.491). Compared with ≤10 km, adjusted hazard ratios were 1.16 for 10–20 km, 0.96 for 20–40 km, 1.02 for 40–60 km and 1.06 for >60 km. The continuous estimate was 1.02 per 25 km (95% confidence interval 0.96–1.09; P =0.452). Distance estimates remained materially unchanged in clinical complete-case and recent-period sensitivity analyses. No formal evidence of distance-effect heterogeneity by diagnosis was detected (interaction P =0.111). Among patients who received non-palliative radiotherapy in this compact referral area, residential distance was not detectably associated with overall survival after adjustment for available measured variables and diagnosis-stratified baseline hazards. These findings do not evaluate geographic access before radiotherapy receipt and do not exclude moderate effects among long-distance patients.