Silvio Giancarlo Bruni BSc, Presenter: Nothing to Disclose

Ravi Menezes PhD, Abstract Co-Author: Nothing to Disclose

Eric Steven Bartlett MD, MPH, Abstract Co-Author: Nothing to Disclose

Eugene Yu MD, Abstract Co-Author: Nothing to Disclose

1) Determine the discrepancy rates in preliminary radiology residents’ interpretations of neuroradiologic CT and MR imaging studies, and quantify any potential or realized effect on patient care. 2) Determine the impact of resident, study and patient variables on discrepancy rate to identify key repetitive patterns of misinterpretation for purposes of improved resident education.

Retrospective chart review of all head, neck and spine imaging studies conducted at our university-affiliated health network from Jan. 1, 2008, through Dec. 31, 2009 revealed 5666 patients with preliminary interpretations made by 57 different on-call residents. Preliminary interpretations were assessed for concordance with interpretations of the final attending neuroradiologist by manual report comparison. For all discrepant interpretations, electronic patient records were investigated for realized or potential impact on patient care. Discrepancies were deemed major in the presence of negative impact on patient care or clinical outcome. Cross-tabs and binomial logistic regression were used to investigate impact of resident, study and patient variables on discrepancy rate.

The overall discrepancy rate for all included studies was 8.0%, with a minor discrepancy rate of 6.8% and major discrepancy rate of 1.2%. Concordance was highest in interpretations of head CT, with minor and major discrepancy rates of 6.1% and 1.3%, respectively, followed by Head and Neck CT, Spine CT, and Brain MR. Spine MR studies demonstrated the poorest concordance, with minor and major discrepancy rates of 11.6% and 1.8%, respectively. Regression analysis revealed a number of factors correlated significantly with discrepancy level, including level of resident training; patient status; time of preliminary interpretation; and presence of acute or remote findings noted in the final report.

Observed discrepancy rates for neuroradiologic imaging studies were comparable to other previously conducted large-scale prospective studies. Our multivariate analyses suggest that clinical outcome is contingent not only on level of interpreter training, but also the presence of acute imaging findings and the effects of "reader fatigue".

Resident preliminary report discrepancy rate is greatly dependent on study type, presence of acute findings and hour; factors to be appropriately considered during resident evaluation or remediation.

Bruni, S, Menezes, R, Bartlett, E, Yu, E, Retrospective Analysis of Factors Involved in Discrepancy between Attending and Preliminary Radiology Resident Interpretations of Overnight and Emergent Neuroradiologic Imaging Studies.  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11003283.html