Do you see what I see? Clinical review of discordant chest x-ray consultant radiologist and radiographer interpretations

Aim
20% observer variation is reported in the literature for chest x-ray (CXR) interpretation. However, not all errors are clinically significant. We examined the clinical significance of discordant radiographer chest x-ray reports.

Methods
100 CXRs were randomly selected from a consecutive series of 1,000 interpreted by a trained reporting radiographer in clinical practice. Three consultant radiologists independently assessed 50 CXRs for agreement with the radiographer report, with 50% overlap to assess inter-radiologist variation. Abnormal cases were categorised as clinically significant/insignificant. Clinical review of discordant cases provided the definitive diagnosis.

Results
99 cases were available for review (40 abnormal cases, 30 significant). Seven CXRs were discordant.

The reporting radiographer over-reported and under-reported two x-rays; one of each was also an error from a radiologist. Mediastinal lymphadenopathy was missed by both the radiographer and one radiologist; granulomas were see on lymph node biopsy. In one case, the radiographer correctly diagnosed heart failure rather than COPD.

Conclusion
Chest x-ray reports occasionally disagree; interpretation of the mediastinum is known to be problematic.. Disagreement between a reporting radiographer and consultant radiologists is similar to inter-radiologist disagreement.