Objective: The radiologic diagnosis of stroke requires accurate detection and appropriate interpretation of relevant imaging findings; both detection and interpretation may be influenced by knowledge of the patient's presentation. In our study, we evaluated the effect of the availability of clinical history on the sensitivity for stroke detection using unenhanced CT and diffusion-weighted MR imaging.
Materials and methods: The records of 733 consecutive patients with a clinically based admission diagnosis of early stroke were reviewed. Among the criteria for inclusion in our study were the availability of an unenhanced CT scan (561 cases) or diffusion-weighted MR imaging examination (409 cases) obtained at admission and a discharge diagnosis indicating whether a patient had actually had a stroke. The radiology requisition forms, available at the time of image interpretation, were classified as either indicating or not indicating a clinical suspicion of early stroke. Sensitivity, specificity, and accuracy of stroke detection were computed, stratified by the presence or absence of an available history indicating suspicion of stroke. Results were compared using the Fisher's exact two-tailed test.
Results: Unenhanced CT sensitivity was 52% (specificity, 95%) for the suspicion-of-stroke group and 38% (specificity, 89%) for the no-suspicion-of-stroke group (p = 0.008). Diffusion-weighted MR imaging sensitivity was 95% (specificity, 94%) for the suspicion-of-stroke group and 94% (specificity, 98%) for the no-suspicion-of-stroke group (p = 0.822).
Conclusion: Availability of a clinical history indicating that early stroke is suspected significantly improves the sensitivity for detecting strokes on unenhanced CT without reducing specificity. In contradistinction, the availability of such a history did not significantly improve the sensitivity for detecting stroke using diffusion-weighted MR imaging. Whenever possible, relevant clinical history should be made available to physicians interpreting emergency CT scans of the head.