(SLC-038) Diagnosis Patterns of Retinal Vasculitis in Real-World Data

Background: Retinal vasculitis (RV) is a rare, vision-threatening ocular condition with acute onset due to infectious and non-infectious causes. RV is typically diagnosed by an ophthalmologist using fluorescein angiography (FA), the diagnostic gold standard. Identifying incident RV cases in secondary data sources is challenging due to diagnostic codes capturing disease recurrence, patient medical history (“carry-over diagnoses”), and codes linked to periodic monitoring for vascular leakage and visual recovery.

Objectives: To describe patterns of RV diagnoses and diagnostic and provider characteristics associated with RV diagnoses in health claims and electronic health records (EHR) data.

Methods: This cohort study used 2 data sources: a general claims database (Komodo) and ophthalmology-specific EHR (Intelligent Research in Sight, IRIS) from 2017 to 2024. The study included patient-eyes diagnosed with incident RV (index date) from January 1, 2019, to December 31, 2021. Patient-eyes were required to have a 24-month pre-index period with wet age-related macular degeneration or diabetic retinopathy but without an RV diagnosis, and a 24-month post-index period. We examined RV diagnosis patterns per eye (e.g., median number of diagnoses, duration from the first to last RV diagnosis, interval between consecutive RV diagnoses). Diagnostic and provider characteristics included FA use, diagnosing provider specialty, and diagnosis position (with primary position indicative of the principal diagnosis).

Results: We identified 290 eyes (mean age: 61±14 years, female: 59%, median follow- up: 40 months) in Komodo and 273 eyes (mean age: 68±14 years, female: 60%, median follow- up: 45 months) in IRIS. The median RV duration was 4 months in Komodo and 27 months in IRIS, with a median of 3 RV diagnoses per eye in Komodo and 8 in IRIS. The median interval between consecutive RV diagnoses per eye was 74 days in Komodo and 61 days in IRIS, with ~50% having ≥1 interval over 6 months and ~20% over 12 months. In Komodo, nearly 75% of all RV diagnoses were made by ophthalmologists and 61% (68% in ophthalmology visits vs. 42% in non-ophthalmology visits) of RV diagnoses were in the primary position. FA use within ±7 days of RV diagnosis in Komodo was 24% (28% in ophthalmology visits vs. 9% in non-ophthalmology visits) and 60% in IRIS.

Conclusions: Persistent RV diagnoses over prolonged periods with long intervals between consecutive diagnoses suggest the need for extended pre-index washout periods to ensure the exclusion of a carryover diagnosis or recurring RV cases. Additionally, the lower use of FA and the one-fourth of diagnoses made by non-ophthalmologists in claims versus an ophthalmology-specific EHR suggest the need for a more refined case definition and validation of RV when using claims data.