Abstract TMP49: Social Network Simulation Identifies Persistent Racial Disparities of Delay to Hospital in Acute Ischemic Stroke Patients

Abstract

Introduction: Delayed arrival to the hospital remains the major reason for non-use of stroke therapies. Minority patients have longer delays that have not been adequately understood nor acted upon. Social context plays a key role, because most strokes occur in front of witnesses who influence decision-making. We created a social network simulation to understand the interpersonal factors that influence decision-making, particularly in minority patients. Methods: We developed an agent-based computer simulation of individual traits and social contextual factors that contribute to decision-making. The inputs into this parsimonious model were based on the largest empirical studies (e.g., Get With The Guidelines) and included: stroke severity, use of Emergency Medical Services (EMS), race, and social network size. The model outputs were the number of stroke patients who arrived at the hospital early (≤3 hours) and late (>3 hours). For each run of the model, 1,000 stroke patients decided to go to the hospital early or late based on individual and social contextual factors. Two sample t-tests were used to compare means between white and non-white patients. Results: In 1.03 million simulations, the model reproduced observed population trends of delay. The overall mean percent of early arrivers was 25.2% (SD 0.02), which matched national metrics showing good calibration of the model. Race predicted delay and modified the relationship of the main effects. 17.9% of non-white patients arrived early compared to 28.3% white patients (p<0.0001) (Fig 1-A). 17.6% of non-white patients with moderate strokes arrived early compared to 39.8% of white patients (p<0.0001) (Fig 1-B). 18.8% of non-white patients with a social network size of 10 arrived early compared to 31.5% of white patients (p<0.0001) (Fig 1-C). Conclusion: A social network simulation reproduced persistent racial disparities of delayed arrival allowing for novel interventions to be tested on this platform.