Request form history, clinical indication, and yield of brain magnetic resonance studies

To investigate whether improved clinical history allows the radiologist to better predict the pretest probability of obtaining a positive or negative result from a magnetic resonance (MR) examination.

Six neuroradiologists prospectively reviewed 100 consecutive requests for brain MR examinations and sequentially assessed 1) quality of written history, 2) degree of indication for requested study, and 3) any pertinent new information found during chart review that may have altered the degree of indication. MR yield was correlated with the degree of indication assessed before and after chart review.

Most request form histories were judged as poor (63%), and chart review reduced the overall indications for MR examinations, as there was a tendency for high-indication requests to migrate to the low-indication category. Based on request form history alone, the yields for low- and high-indication studies were 13% and 37%, respectively. Correlations between MR yield and indication after chart review improved significantly (P < 0.05) with 2% and 61% for low and high indications, respectively. Sensitivity and specificity for a positive MR yield were 71% and 62%, respectively, for the indication judged by the request history alone, and 96% and 80%, respectively, after chart review. Positive and negative prediction rates were 37% and 87%, respectively, for the indication judged by the request history alone, and 61% and 98%, respectively, for the indication judged after chart review.

Based on our limited data, most request form histories were inadequate, and essential information available in the chart before MR examinations was frequently missing from the request forms. When adequate information was provided, the indication for the studies as judged by the radiologists predicted the MR yield more accurately, particularly for those requests with low indication. Therefore, our study suggests that MR imaging (MRI) may be used more effectively when pertinent clinical history is available. However, our study is limited and further studies are needed to confirm our results.