Computed tomography evaluation in acute stroke: retrospective study

ACF Hui, CY Man, ASY Tang, KM Au-Yeung
Department of Medicine and Therapeutics, The Chinese University of Hong Kong, Prince of Wales Hospital, Shatin, Hong Kong

OBJECTIVE. To determine the accuracy with which ‘frontline’ hospital doctors interpret computed tomography brain scans.

DESIGN. Retrospective study.

SETTING. University teaching hospital, Hong Kong.

PARTICIPANTS. Medical and emergency room doctors.

MAIN OUTCOME MEASURES. Accuracy in correctly identifying features of acute stroke on 18 computed tomography brain scans.

RESULTS. Computed tomography brain scan images showing easily detectable haemorrhage and infarct were identified in 91% and 90% of scans, respectively; but difficult-to-interpret scans with subtle features of haemorrhage or infarct were only correctly identified in 46% and 45% of readings, respectively. More experienced doctors did not perform better than junior doctors (P=0.69; 95% confidence interval, -1.84 to 2.73) and the mean total score for doctors from the emergency department did not differ significantly from that of doctors from the medical department (P=0.57; 95% confidence interval, -2.98 to 1.67).

CONCLUSION. Early signs of infarct and small bleeds on computed tomography brain scans are not well recognised by doctors, regardless of clinical exposure or seniority. Ineligible patients may be treated with thrombolytic therapy as a result of such computed tomography scan misinterpretation.

Hong Kong Med J 2002;8:177-80

Key words: Cerebral infarction; Cerebrovascular accident; Tomography, X-ray computed