A methodologic assessment of studies comparing magnetic resonance imaging and computed tomography of the brain

To evaluate the methods used in studies comparing magnetic resonance imaging (MRI) and computed tomography (CT) of the brain, we reviewed English language articles published between 1981 and early 1986. Of 83 studies comparing the accuracy of MRI and CT in 10 or more patients, few met methodologic standards for evaluations of the efficacy of new diagnostic tests. Limitations were found in how study populations were assembled, in test performance and interpretation, and in analysis of study findings. Greater attention to design and execution of comparison studies could overcome these limitations and help demonstrate MRI's true diagnostic value.