Fool me twice: delayed diagnoses in radiology with emphasis on perpetuated errors

We hypothesized that delayed diagnoses in radiology are not recognized on subsequent radiologic examinations because of multiple types of errors.

Six hundred fifty-six radiologic examinations with delayed diagnoses were collected from July 1, 2002, to January 31, 2010. Each case was reviewed by two radiologists together, and the diagnostic errors were classified according to our modified scheme with consensus between the radiologists.

There were a total of 1269 errors. The range of days elapsed from the initial error in interpretation to the correct diagnosis was 0-4611 days, with an average of 251 days. The percentage for each type of error was 0.9% (n=11) for type 1, 9% (n=110) for type 2, 3% (n=39) for type 3, 42% (n=535) for type 4, approximately 0% (n=1) for type 5, 2% (n=29) for type 6, 5% (n=59) for type 7, 2% (n=20) for type 8, 7% (n=92) for type 9, 22% (n=288) for type 10, 0.5% (n=6) for type 11, and 6% (n=79) for type 12. The correct diagnoses were not recognized on subsequent radiologic examinations in 196 of 656 cases (30%).

Delayed diagnoses were not recognized on subsequent radiologic examinations in about one third of the cases. The most common types of error were underreading, satisfaction of search, faulty reasoning, and location of the finding.