Diagnostic errors in clinical FDG-PET/CT

Abstract

This retrospective study included 4,099 consecutive clinical FDG-PET/CT scans with corresponding reports that were made at a tertiary care center in an 18-month period. FDG-PET/CT reports were scrutinized for the presence of an addendum enclosing a diagnostic error. Results: 90 of 4,099 FDG-PET/CT reports (2.2%) contained an addendum enclosing a diagnostic error. The distribution of perceptual and cognitive errors among these 90 diagnostic errors was 54 (60.0%)/36 (40.0%). On multivariate logistic regression analysis, only low-dose FDG-PET/CT combined with concomitantly acquired and interpreted full-dose contrast-enhanced CT remained as significantly and independently associated with the presence of a diagnostic error, relative to low-dose FDG-PET/CT without concomitantly acquired and interpreted full-dose contrast-enhanced CT (odds ratio: 2.79 [95% confidence interval: 1.61-4.85], P < 0.001). Patient age, gender, hospital status, indication for FDG-PET/CT scanning, single vs. double reading (i.e. two medical imaging specialists), reader experience, and reading by a nuclear medicine physician only vs. reading by both a nuclear medicine physician and a radiologist, were not significantly and independently associated with the presence of a diagnostic error. Conclusion: Diagnostic errors in clinical FDG-PET/CT based on addenda to the original report are relatively infrequent, though certainly non-negligible. Perceptual errors are slightly more frequent than cognitive errors. The availability of a concomitantly acquired and interpreted full-dose contrast-enhanced CT seems to increase diagnostic error rate. These data can be used for quality improvement and benchmarking purposes.

Deep learning for [18F]fluorodeoxyglucose-PET-CT classification in patients with lymphoma: a dual-centre retrospective analysis

“…Alotaibi and colleagues analysed 4099 PET-CT reports generated over an 18-month period at a tertiary care centre, and found that 2·2% contained an addendum that revealed a retrospectively discovered diagnostic error. 19 The true error rate for PET-CT reports by physicians, which would also include errors that go unnoticed, remains unknown, but is probably higher, given that for radiology tests in general, a human error rate of 3–5% has been estimated. 20 To achieve comparable sensitivity levels (eg, 95%) with our model, modification of the cutoff value between positive and negative scans, at the expense of specificity, might be an option ( table 2 ).…”

Learning from error: a nuclear medicine events and learning meeting

“…One study of incidents in radiology and nuclear medicine by Martin found that common errors in nuclear medicine include overexposure of radiation to patients, contamination events, and failure of management of radiopharmaceuticals [3]. Another study by Alotaibi et al [4]. found a diagnostic error rate of 2.2% in 4099 fluorodeoxyglucose PET-computed tomography reports.…”