Reference ranges identified for discrepancy rates in adult CT studies

Discrepancy rates for reporting of adult CT studies and potential useful reference ranges have been identified, according to a meta-analysis study published in the March issue of Radiology.

Nailing down a baseline discrepancy rate for imaging examination interpretation is vital for assessing radiologist skills. Lead author Mark Z. Wu, MSc, of the Ottawa Hospital Research Institute in Canada, and colleagues sought to search the medical literature to assess the discrepancy rate when interpreting CT studies in adult patients, as well as to establish whether the rate differs by basis of body region or radiologist training.

Using two databases, the researchers searched from 1946 to June 2012 for studies that met the inclusion criteria. Data were extracted from 58 studies and total and major discrepancy rates were then calculated with meta-analysis.

Results of the analysis revealed a pooled total discrepancy rate of 7.7 percent and a major discrepancy rate of 2.4 percent. The pooled major discrepancy rate was comparable for staff, whose rate was 2.9 percent, and residents, whose rate was 2.2 percent.

The pooled major discrepancy rate for head CT was 0.8 percent and 0.7 percent for spine CT. These findings were lower than those for chest and abdominal CT, which were 2.8 percent and 2.6 percent, respectively.

Reference radiologists who were not blinded to the initial report had a 2 percent major discrepancy rate, which was much lower than that of 12.1 percent for those who were blinded.

“This study advances our understanding of discrepancy rates for reporting of adult CT studies, with potentially useful reference ranges identified in the subgroup analyses on the basis of body region scanned; however, considerable heterogeneity that is only partially explained by subgroup analysis signifies that further research to refine our understanding of this important area is necessary—particularly regarding the question of blinding of the reference radiologist,” wrote Wu and colleagues.