Background:
Prompt signs and symptoms recognition and intervention are essential to achieve the best outcome after stroke. Stroke codes were developed to expedite assessment and treatment. Their optimal use requires accurate identification of stroke patients. In order to improve diagnostic accuracy in our institution, we analyzed the predictive value of individual stroke signs and symptoms in patients in whom stroke codes were activated from the emergency department (ED) by physicians and nurses and from inpatient wards by nurses, residents and hospitalists.
Methods:
We retrospectively analyzed 501 consecutive stroke codes in our stroke log from May 2013 to May 2015. Age, gender, presenting signs and symptoms, medical history and final diagnosis were assessed. Patients were classified as stroke (ischemic and hemorrhagic) or non-stroke based on the final impression after the completed work-up. X2 statistic was utilized to assess associations.
Results:
Overall, 202 (40.3%) patients were classified as stroke and 299 (59.7%) non-stroke. 78% of stroke codes were activated from ED and 22% from the inpatient wards. Unilateral limb weakness, aphasia and facial weakness were associated with stroke (p<0.05) with PPVs of 0.57 (95%CI 50-64%), 0.56 (43-68%), 0.51 (43-60%), respectively. Altered mental status (AMS) and sensory symptoms were associated with non-stroke (p<0.05). The PPV and NPV for stroke were 0.21 (95%CI 13-31%) and 0.55 (50-60%) for AMS respectively and 0.25 (14-39%) and 0.58 (43-63%) for sensory symptoms. Location of the stroke code (ED or inpatient ward) did not impact the results.
Conclusion:
Previous studies, based on evaluation of acute stroke by paramedics and ED physicians, demonstrated that some signs or symptoms are more likely to be present in patients experiencing acute stroke. In our experience, unilateral limb weakness, aphasia, and facial weakness as identified by diverse provider disciplines and experience levels are associated with a final diagnosis of acute stroke. However, isolated altered mental status or sensory symptoms seldom result in a final diagnosis of stroke. These data can assist healthcare providers, to more accurately identify stroke patients, thus improving outcomes as well as resources utilization.