Objective: To assess clinicians' agreement on how they interpret lung scan reports with regard to the diagnosis of pulmonary embolism.
Design: In this prospective study, nuclear medicine physicians provided two types of reports for each lung scan: a routine descriptive report and a short form with a standardized conclusion on the likelihood of pulmonary embolism: "high probability," "no conclusion," and "diagnosis excluded." Three independent blinded senior clinicians reviewed all routine reports and chose one of the following conclusions: "high probability," "no conclusion," or "diagnosis excluded."
Setting: An acute care teaching hospital near Paris.
Subjects: Eighty-two lung scans were studied.
Main outcome measurements: Inter-clinician agreement and agreement between clinicians' conclusions and the nuclear medicine physicians' standardized reports were analyzed using the kappa index.
Results: The distribution of the clinicians' conclusions from routine reports strongly differed (p < 0.001). Agreement among the three clinicians was observed in 40.2% of the routine reports, and the inter-clinician agreement was poor to moderate (kappa range, 0.28 to 0.52). A complete agreement among the three clinicians and the nuclear medicine physicians' standardized conclusions was observed for 32.9% of the reports. The agreement between each clinician and the standardized conclusions was also poor to moderate (kappa range, 0.32 to 0.55).
Conclusion: Reading the same routine reports, clinicians reached different conclusions. Furthermore these differed greatly from the nuclear medicine physicians' standardized conclusions. These results support the notion that physicians should be given standardized reports.