Introduction: The risk of ischemic stroke is significantly higher in pregnancy. However, the neurologic symptoms that accompany stroke can be mimicked by other conditions. Because ischemic stroke has the potential for acute treatment, it is important to identify these patients quickly.
Methods: A retrospective chart review was performed on pregnant patients from 2009-2014 using ICD-9 codes for stroke or other related vascular diseases. The diagnosis was based on the neurologist evaluation and neuroimaging. We identified candidate variables with p<0.2 in univariable analysis and derived a predictive model using interactive variable selection, then fit logistic regression models to the outcome of ischemic stroke.
Results: Thirty-eight patients met inclusion criteria, of which 13 had ischemic stroke (Table 1). Mimic diagnoses included: venous sinus thrombosis (8), intracerebral hemorrhage (2), complicated migraine (7), non-epileptic seizure (3), seizure (2), transient ischemic attack (1), and conversion disorder (2). The best model to predict ischemic stroke included age greater than 29 years and history of 1st degree relative with stroke (Figure 1).
Conclusion: Our scoring system is a promising hypothesis-generating concept that requires validation in a larger cohort of pregnant patients. Table 1: Baseline characteristics of patients with ischemic stroke versus mimics and p-value for difference (using Student’s t-test or the Chi-squared test).
Figure 1: Comparison of the area under the Receiver Operating Characteristic curve for Model 1 (current smoking and age>29), Model 2 (migraine with aura and age>29), Model 3 (family history of stroke and age>29).