Clinical yield of computed tomography brain scans in older general medical patients

OBJECTIVES: To evaluate the clinical yield of computed tomography (CT) brain scans in a prospective cohort of older patients admitted to the general medicine service. DESIGN: Nested cohort study of 117 subjects enrolled in previous prospective cohort study of 919 subjects. SETTING: University-affiliated teaching hospital. PARTICIPANTS: Hospitalized general medical patients aged 70 and older who received one or more brain CT scans during their hospital stay. MEASUREMENTS: Review of medical records and interpretation of the first brain CT scan in these 117 patients for indications for ordering scans and clinically significant brain abnormalities. Medical records of patients with brain CT scans with abnormalities were reviewed for 2 weeks after the scan for changes in medical management resulting from scan findings. Three independent reviewers adjudicated the presence of abnormalities and resulting treatment changes. RESULTS: Of the 117 brain CT scans, 32 (27%) were ordered to exclude intracranial hemorrhage, 30 (26%) to exclude cerebrovascular accident (CVA), 16 (14%) for falls, 15 (13%) for syncope, seven (6%) to exclude subdural hemorrhage, five (4%) for mental status change, and 12 (10%) for other reasons. Of the 117 brain CT scans, 29 (25%) had abnormalities, including acute CVA or hemorrhage, old CVA, meningioma, and other abnormalities. Only 10 (9% of all scans, 34% of abnormal scans) resulted in treatment changes (including consultations, further imaging, stroke evaluation, and drug changes). The presence of focal neurological deficits was significantly associated with treatment changes after CT scans (odds ratio=13.2, 95% confidence interval=1.7-161.5). CONCLUSION: These results suggest that the overall clinical yield of brain CT scans in unselected older hospitalized patients is low. Targeting scans toward patients with new focal neurological deficits will help to improve clinical yield.