To evaluate junior neurology resident diagnostic accuracy among patients requiring intensive care unit (ICU) management, to help guide educational initiatives for resident assessment of high-acuity cases. 

Through the Close the Loop Project at Mount Sinai, we recorded residents’ initial diagnostic impression in cases presenting in morning report from July 2010 to June 2016. We compared residents' initial impression to the final diagnosis and characterized diagnostic errors. In this set of analyses, we stratified diagnostic accuracy according to disposition, critical care needs, and high-acuity diagnoses. Significance was determined using analysis of variance models.

1301 cases were included, with a previously reported overall accuracy of 64.0%. Patient disposition was significantly associated with resident accuracy (p<0.01). 177 patients did not require hospital admission (accuracy 79.0%). Of admitted patients, 348 were admitted to the neurology service (accuracy 64.7%), 308 to the stroke unit (accuracy 65.3%), and 328 to a primary medical/surgical team (accuracy 64.9%). 140 patients required ICU care and/or neurosurgical intervention, with significantly lower resident accuracy (55.3%, p=0.03). While ischemic strokes and seizures were appropriately identified by residents (sensitivity 79.9% and 79.6%), residents were more likely to misdiagnose intracranial hemorrhages (sensitivity 23.8%) before neuroimaging was obtained.

This analysis of our clinical acumen assessment database suggests that junior neurology resident accuracy varies with level of care required, with lower diagnostic accuracy in cases initially requiring critical care and urgent neurosurgical intervention. Our analyses explore the nature of diagnostic patterns among these patients, which can help guide educational/supervision approaches for these high-acuity clinical situations.