Abstract Body: Introduction
Stroke misdiagnosis is not uncommon and Black people have a higher odds of stroke misdiagnosis than White people. Little is known about the reasons for misdiagnosis or how diagnostic pathways lead to error.

Hypotheses
To explore possible explanations for race differences in stroke diagnosis, we examine two hypotheses. First, to understand where diagnostic pathways may err we quantified, whether race predicted whether receipt of initial imaging (ie. CT/MRI), hypothesizing that black people receive less imaging. Second, amongst individual that receive imaging, we hypothesized that Black people will be less likely to receive stroke diagnoses or be admitted to the hospital.

Methods
We used data from the National Hospital Ambulatory Medical Care Survey (NHAMCS) from 2011-2021. Our primary study population was adults with any neurologic chief complaint, defined using reason for visit (RFV) data. Our primary exposure was Black race. For each outcome, we built unadjusted, minimally adjusted, and fully adjusted survey-weighted logistic regression models to predict outcomes: receipt of imaging, stroke diagnoses and hospital admission. The minimally adjusted model added age, sex, and insurance to race/ethnicity. The fully adjusted model added vital signs, arrival by ambulance, triage level, episode of care, vascular risk factors, prior vascular disease and RFV-based characterization of stroke probability.

Results
The mean age was across the ~146,000 emergency room visits 47.8 ± 19.7 and black people comprised 24.9% of the sample. Amongst individuals with a neurologic RFV, black people were less likely to get a CT (OR 0.79, 95% CI 0.72-0.88), MRI (OR 0.64, 95% CI 0.47-0.89), or any imaging (OR 0.78, 95% CI 0.71-0.86 ) in the unadjusted model. These effects persisted or were strengthened in the fully adjusted model: Black race-imaging association (OR 0.67, 95% CI 0.45-0.99). Conditioned on receiving imaging after full adjustment, there was no association between Black race and stroke diagnoses (OR 1.1, 95% CI 0.63-1.9) or hospital admission for stroke (OR 1.19, 95% CI 0.46-3.08).

Conclusion
We found that Black people were less likely to receive imaging when presenting with neurologic reasons for visit, but that conditioned on receiving imaging, there were no differences in stroke diagnosis or hospitalization. This suggests that race differences in stroke diagnosis likely occur at or prior to the time of initial imaging selection.