Should registrars be reporting after-hours CT scans? A calculation of error rate and the influencing factors in South Africa

Abstract

There is a heavy reliance on registrars for after-hours CT reporting with a resultant unavoidable error rate. To determine the after-hours CT reporting error rate by radiology registrars and influencing factors on this error rate. A 2-month prospective study was undertaken at two tertiary, level 1 trauma centers in Johannesburg, South Africa. Provisional CT reports issued by the registrar on call were reviewed by a qualified radiologist the following morning and information relating to the number, time and type of reporting errors made as well as the body region scanned, indication for the scan, year of training of the registrar, and workload during the call were recorded and analyzed. A total of 1477 CT scans were performed with an overall error rate of 17.1% and a major error rate of 7.7%. The error rate for 2nd, 3rd, and 4th year registrars was 19.4%, 15.1%, and 14.5%, respectively. A significant difference was found between the error rate in reporting trauma scans (15.8%) compared to non-trauma scans (19.2%) although the difference between emergency scans (16.9%) and elective scans (22.6%) was found to be not significant, a finding likely due to the low number of elective scans performed. Abdominopelvic scans elicited the highest number of errors (33.9%) compared to the other body regions such as head (16.5%) and cervical, thoracic, or lumbar spine (11.7%). Increasing workload resulted in a significant increase in error rate when analyzed with a generalized linear model. There was also a significant difference noted in the time of scan groups which we attributed to a workload effect. Missed findings were the most frequent errors seen (57.3%). We found an increasing error rate associated with increasing workload and marked increase in errors with the reporting of abdominopelvic scans. There was a decrease in the error rate when looking an increasing year of training although this there was only found to be significant difference between the 2nd and 3rd year registrars.