Abstract Number: Esoc2026a2034 Long-Term Trends in Stroke Recurrence and Sociodemographic Inequalities in a Diverse Population: 30 Years of the South London Stroke Register

Abstract Background and aims Stroke survivors remain at high risk of recurrence. Long-term recurrence trends and associations with sociodemographic factors (age, ethnicity, socioeconomic status) are poorly understood. This study investigates how these are associated with recurrence over 30 years. Methods Participants from the South London Stroke Register (N=7,836) with a first-ever stroke between 1995 – 2025 were analysed. The main outcome was first recurrent stroke. Cox proportional hazards and Generalized Estimating Equation models were used to assess associations with predictor variables. Results There were a total of 7930 participants, with a median age of 70.4 58.9-80.4 years. 4733 (60%) were white, 1167 (15%) were of Black Caribbean ethnicity, 1051 (13%) were of Black African ethnicity, and 778 (10%) identified as others. 4057 (51%) were of low IMD. 939 participants (11.8%) experienced their first recurrence over a median follow-up of 3.0 years. After accounting for mortality, recurrence risk was highest in 1995-1999, declined significantly (OR=0.27 0.12 - 0.61) in 2000-2004, and then only modestly for the next 20 years. Older age (HR=1.02 1.02 - 1.03) and low IMD (high vs. low IMD; HR = 0.75 0.59 - 0.96) were key predictors of recurrence. Conclusions Recurrence risk declined significantly early on, but thereafter improvements were marginal for two decades. Age and socioeconomic deprivation are key risk factors. Further risk reductions require shifts towards policies targeting high-risk subgroups. Conflict of interest Evelyn Lim: nothing to disclose. Aicha Goubar: nothing to disclose. Eva Emmett: nothing to disclose. Camila Pantoja-Ruiz: nothing to disclose. Ajay Bhalla: nothing to disclose. Ismail Ismail: nothing to disclose. Iain J. Marshall: nothing to disclose. Matthew D. L. O’Connell: nothing to disclose Figure 1 - belongs to Results