Quality of Interpretation Of Computed Tomography Images In The Emergency Department

Abstract

Patient care is increasingly dependent on emergency department services. Teleradiology provides a method of supplementing radiology services. We conducted a prospective study to evaluate the accuracy of interpretation for cranial computed tomography (CT) images transmitted by Picture Archives Communication System (PACS), with the image viewed on a computer screen. In addition, incidents of misinterpretation leading to an alteration in patient care were reviewed. Five hundred and thirty-four patients undergoing cranial CT scanning in the emergency department (ED) from December 1995 to January 1996 were reviewed. The PACS transmitted 483 patients’ cranial CT images successfully. The interpretations of the PACS transmitted images, by emergency physicians (EPs), were compared with those of the original films, by radiologists. We then asked the EPs to read the original CT films in a randomly selected 25% (121 out of 483) of successfully transmitted cases, in order to compare differences in image reading between original films and PACS images. The leading three indications for CT scanning were neurologic focal signs (200, 37.5%), altered mental status (144, 27.0%), and trauma (120, 22.5%). The other indications were headache (41, 7.7%), seizure (38, 7.1%) and dizziness, vertigo and/or vomiting (30, 5.6%). The CT scans were reported to be positive by radiologists in 403 cases (75.5%). The top ten abnormalities were infarction (53.1%), subcortical arteriosclerotic encephalopathy, SAE (22.1%), parenchymal hemorrhage (16.6%), brain atrophy (14.9%), calcification (12.9%), cerebral edema (8.4%), scalp hematoma (7.7%), midline shift (7.4%), mass (6.7%), and subarachnoid hematoma (6.2%). Non-concordance between the CT interpretations by the EPs and radiologists were found in 132 cases (27.3%). These non-concordances included 6 major false negatives (1.2%), 96 minor false negatives (19.9%), 17 minor false positives (3.5%), 13 minor false negative + minor false positives (2.7%), and no major false positives. Most of the non-concordant interpretations by EPs were lacunar infarction, calcification, SAE, sinusitis and brain atrophy. The six cases with major false negative included 3 infarctions, 2 masses, and I subdural hematoma, however, no patient was managed inappropriately, and none had an adverse outcome. The same results were achieved in 76 cases (62.8%) when the EPs read the original CT films, as compared to reading the PACS transmitted images. The EPs reported more .findings in 15 cases (12.4%), and less in the other 30 cases (24.8%). The time to interpretation post CT scan was 2.1 ±3.0 hours for EPs, and 20.3 ±22.3 hours for radiologists (p＜0.001). Misinterpretation of cranial CT scans by EPs is of potential clinical concern, however, no resultant clinical errors were found in this study. We recommend that formal training in CT interpretation be included in residency training and continuing medical education programs for EPs, to ensure important errors are not made during the acute early phase of care.