Ocular cancer and inpatient thrombotic phenotypes: Associations with retinal vascular occlusions and venous thromboembolism in the 2018-2022 National Inpatient Sample.

Abstract

e23424 Background: Cancer-associated thrombosis is common, but ocular malignancy may have distinct local and systemic vascular effects. Ocular tumors and ocular radiotherapy can injure retinal microvasculature, yet population-level evidence linking ocular cancer to retinal occlusion and systemic VTE is limited. We evaluated associations between ocular cancer, venous thromboembolism (VTE), and retinal vascular occlusions. Methods: Retrospective cross-sectional analysis of 2018-2022 National Inpatient Sample (NIS) discharges (weighted U.S. estimates). Exposure was ocular cancer (ICD-10-CM C69). Outcomes were VTE (I26/I82) and retinal vascular occlusions (H34), subclassified as retinal vein occlusion (RVO) and retinal artery occlusion (RAO). Propensity-score matching balanced demographics, hospital characteristics, and comorbidity burden (Charlson Comorbidity Index, CCI). Post-match associations were estimated using negative binomial regression with log(length of stay) offset to obtain crude IRR and adjusted IRR (aIRR) with 95% confidence intervals (CI), incorporating NIS survey weights and robust standard errors. Length of stay and total charges were assessed with log-linear regression; effects are reported as percent change. Analyses were performed in Stata 18. Two-sided p < 0.05 was significant. Results: Among 33,905,592 weighted hospitalizations, 3,010 (0.009%) had ocular cancer. Matching improved balance (age standardized difference 32.1% to 4.1%). Unmatched event rates were VTE 2.79% vs 2.74% (p = 0.624), any RVO 0.158% vs 0.013% (p < 0.001), and any RAO 0.071% vs 0.021% (p = 0.155). After matching/adjustment, ocular cancer was associated with higher RVO (aIRR 12.16, 95% CI 3.49-42.34; p < 0.001) and lower VTE (aIRR 0.67, 95% CI 0.53-0.84; p < 0.001); RAO was not significant (aIRR 3.39, 95% CI 0.63-18.26; p = 0.155). RAO estimates were imprecise. In CCI-stratified analyses, RVO remained elevated in low/medium/high CCI (aIRR 8.52 [1.85-39.21], 13.45 [3.25-55.62], 11.21 [2.15-58.33]); VTE reduction was greatest in low CCI (aIRR 0.58 [0.36-0.94]). Ocular cancer hospitalizations had longer length of stay (+44.5%; coef 0.368 [0.330-0.405]) and higher charges (+126.4%; coef 0.817 [0.785-0.849]) (both p < 0.001). Conclusions: In national inpatient data, ocular cancer was strongly associated with retinal venous occlusion but lower coded VTE, diverging from typical cancer-associated thrombosis patterns. Findings are compatible with predominant local ocular venous vulnerability (tumor-related and/or treatment-related) rather than systemic hypercoagulability, but outpatient VTE and treatment exposures are not captured. Given discharge-level ascertainment and limited tumor detail; longitudinal studies incorporating subtype, stage, treatment exposure, and outpatient events are warranted.