Published online before print February 2, 2012, doi: 10.3174/ajnr.A3023
AJNR 2012 33: E42

D. Friedman^a
aDepartment of Radiology
Thomas Jefferson University Hospital
Jefferson Medical College
Philadelphia, Pennsylvania

I read with interest the article titled “Interpretation Errors in CT Angiography of the Head and Neck and the Benefit of Double Reading” by Lian et al¹ in the December issue of the American Journal of Neuroradiology. I think that the analysis offered by the authors would be even more useful if they had provided the following additional information:

• 1) The phenomenon of “satisfaction of search” is known to be responsible for perceptual errors in radiology. For example, if a patient has an intracranial CTA for left Sylvian fissure predominant subarachnoid hemorrhage and an MCA aneurysm is identified, it is probably more likely that an additional small aneurysm or completely unrelated finding will be missed because the radiologist can “relax” now that the patient’s problem has been identified. Conversely, if the expected aneurysm is not found, the radiologist will look very long and hard for any other aneurysm or an alternative explanation for the hemorrhage. Did satisfaction of search play a potential role in the patients with missed findings in this study?

• 2) Given the large volume of information on CTA, the radiologist tends to expend the greatest effort trying to identify the most clinically relevant abnormalities. As such, it would be useful to know whether more vessel occlusions were missed in patients being studied for intracranial hemorrhage and more aneurysms were missed in patients being evaluated for stroke.

• 3) The amount of information in a stroke protocol is approximately double that of either the intracranial or carotid protocols because both the head and neck are studied in the stroke protocol. Was the miss rate higher on these double studies compared with the single studies? The authors also do not provide a breakdown of the number of each type of protocol in the “Results.”

• 4) It would be interesting to know whether the miss rate was higher for studies interpreted “on call,” when the radiologist necessarily works more quickly, compared with studies performed during the day.

• 5) The age of the patients was reportedly recorded, though the information is not provided in the “Results. ” Elderly patients are more likely to have multiple abnormalities, incidental or otherwise. Were there more missed findings in elderly compared with young patients?

• 6) As defined by the authors, the term “significant” discrepancy is somewhat ambiguous. We are not provided with any specific demographic information regarding the patients with missed findings. Is a 1-mm aneurysm in a 75-year-old patient a “significant” finding? Is an A2 occlusion in a 90-year-old patient with subarachnoid hemorrhage a “significant” finding? Can the authors give examples of “minor” discrepancies?

• 7) Five radiologists were the initial readers of the CTA studies, including 3 with a great deal of experience (16–35 years). Was the distribution of errors equivalent among all of the radiologists? Was there a disproportionate number of errors by a particular radiologist, which could have skewed the results unfavorably?

I believe that this additional analysis would strengthen the results of this provocative article.

Reference

1. Lian K, Bharatha A, Aviv RI, et al. Interpretation errors in CT angiography of the head and neck and the benefit of double reading. AJNR Am J Neuroradiol 2011; 32: 2132–35 Abstract/FREE Full Text

Reply

Published online before print February 2, 2012, doi: 10.3174/ajnr.A3042
AJNR 2012 33: E43

S.P. Symons^a, K. Lian^a, A. Bharatha^a and R.I. Aviv^a
aDivision of Neuroradiology
Department of Medical Imaging
University of Toronto
Toronto, Ontario, Canada

Thank you for your interest in our article “Interpretation Errors in CT Angiography of the Head and Neck and the Benefit of Double Reading.”¹ We discuss each of your questions as numbered.

• 1) We agree that the phenomenon of “satisfaction of search” can be responsible for perceptual errors in radiology. We reassessed our data to determine how many of the misses were in the setting of a correctly detected major abnormality that explained the presenting complaint. There were 8 misses on 7 studies in which “satisfaction of search” may have played a role.

• 2) Vessel occlusions and stenoses were only considered a significant miss in the setting of stroke. Therefore, none of these misses occurred in the setting of hemorrhage. Of the 13 missed aneurysms, 7 were missed when the presenting history was rule out aneurysm. The other 6 had presenting complaints of stroke (n = 1), vasculitis (n = 1), stenosis (n = 3), and vascular injury (n = 1).

• 3) For the complete head and neck CTA studies, there were 13 misses on 11 studies. For the intracranial-only CTA studies, there were 13 misses on 9 studies. The difference was not statistically significant.

• 4) There were 5 misses on 3 studies that occurred on call. There were 21 misses on 17 studies that occurred during regular work hours. The miss rate was not higher on call compared with regular work hours.

• 5) The mean age of patients with misses was 65 ± 15 years. The mean age of patients without misses was 60 ± 19 years. The difference was not statistically significant.

• 6) Only findings that we believe required follow-up, were pertinent to the presenting symptom, or required immediate intervention were defined as “significant. ” Examples of minor discrepancies included stenoses unrelated to the presenting symptom and degenerative changes. We included all missed aneurysms, even if the presenting symptom was not subarachnoid hemorrhage, because it was judged that these would require follow-up. Arguably some of these may not be significant.

• 7) The distribution of errors was not equivalent among all of the radiologists, but the difference was not statistically significant. The least experienced radiologist had 5 misses on 4 studies. The second least experienced radiologist had 4 misses on 3 studies. The most experienced radiologist had 6 misses on 5 studies. The second most experienced radiologist had 11 misses on 8 studies. The middle radiologist in terms of experience had no misses. He worked only part-time at the time of the study and reported the fewest number of studies.

Reference

1. Lian K, Bharatha A, Aviv RI, et al. Interpretation errors of CT angiography of the head and neck and the benefit of double reading. AJNR Am J Neuroradiol 2011; 32: 2132–35 Abstract/FREE Full Text