Radiology residents' on-call interpretation of chest radiographs for congestive heart failure

RATIONALE AND OBJECTIVES: This study was designed to evaluate the performance of radiology residents in interpreting emergency department (ED) chest radiographs for congestive heart failure and to characterize the factors associated with a subsequent amended interpretation by an attending radiologist. MATERIALS AND METHODS: We retrospectively reviewed all amended reports for ED chest radiographs between January 2004 and July 2005 and identified those with discrepant interpretations regarding the diagnosis of congestive heart failure. A total of 1.9% (476 of 24,600) of chest radiographs were amended over the study period. Forty-eight patients (75% female, mean age 66 years) whose chest radiograph was amended for the diagnosis of congestive heart failure and were available for review formed the study population. A control group of 35 patients (69% female, mean age 67 years) were individually matched to a convenience subset of patients by age, gender, clinical indication, and radiographic projection. Chest radiographs were in the anteroposterior projection in 62% (30 of 48) of study patients and 60% (21 of 35) of controls. A blinded expert panel of three board-certified cardiothoracic radiologists jointly reviewed each chest radiograph for the presence or absence of congestive heart failure and its specific radiographic findings. RESULTS: The expert panel diagnosed congestive heart failure in 19% (9 of 48) of study patients and in 23% (8 of 35) of controls (P = .65). When present, congestive heart failure was mild to moderate in severity in both the study and control groups (P = 1.00). There was a significant difference in the expert panel agreement between the attending versus the resident interpretation (65% versus 35%, P = .008), for the study group. This resulted in fair agreement (kappa = 0.29) between the expert panel and the attending interpretation and no agreement (kappa = -0.29) between the expert panel and the resident interpretation. In contrast, the expert panel agreed with the joint resident/attending interpretation in 83% (29 of 35) of controls, yielding substantial agreement (kappa = 0.72). CONCLUSION: Interpretation of chest radiographs for congestive heart failure by radiology residents has a low error rate. The majority of chest radiographs with discrepant resident and attending interpretations were portable films of female patients with subtle radiographic findings of congestive heart failure, and were inherently difficult to interpret.