Effect of clinical information in brain CT scan interpretation : a blinded double crossover study

Errors and variations in interpretation can happen in clinical imaging. Few studies have examined the biased effect of clinical information on reporting of brain CT scans. In a blinded double crossover design, we studied whether three radiologists were biased by clinical information when making CT scan diagnosis of the brain. Three consultant radiologists in three rounds with at least a one month interval assessed 100 consecutive cases of brain CT scan. In the first round, clinical information was not available and 100 films without clinical information were given to radiologists. In the second round, the same 100 films were given and true clinical information was available. In the third round, the same 100 films were given and false clinical information was allocated. In 180 cases (60%) the evaluation resulted in the same diagnosis on all three occasions (95% confidence interval (CI): 54.5, 65.5), whereas 120(40%; 95% CI:34.5, 45.5) sets were evaluated differently. 48 cases (16%; 95% CI:11.9,20.1) had discordant evaluation with true and 33 (11%; 95% CI:7.5, 14.5) with false clinical information. Discordance without and with true and false clinical information was 39 (13%; 95% CI:9.2, 16.8). Correct clinical information improves the brain CT report, while the report became less accurate false clinical information was allocated. These results indicate that radiologists are biased by clinical information when reporting brian CT scans