RATIONALE AND OBJECTIVES: Prospective studies of radiologists' interpretations of selected radiographs reported 20-40 years ago indicated error rates of 30% and higher. The authors retrospectively evaluated the interpretations of groups of radiologists and determined a range of rates of disagreement in interpretation. Quality assessment or recredentialing may add to the importance of such studies in the future. MATERIALS AND METHODS: Over a 7-year period, a team of radiologists reviewed imaging interpretations in the radiology departments of six community hospitals. Each review, which lasted about 3 days, included evaluation of the interpretations of a 3%-4% sample of the images read by the radiologists at these hospitals. Reading errors were quantitated and evaluated qualitatively. RESULTS: In a review of over 11,000 images read by 35 radiologists, the authors found a 4.4% mean rate of interpretation disagreement; only one radiologist had a mean rate above 8%. Qualitative analysis of the interpretation errors revealed a mean rate of 3.0% of errors that were considered to be below an acceptable standard of care. Radiologists whose errors included a relatively high proportion of false-positive findings tended to make relatively fewer total errors. CONCLUSION: Rates of disagreement for a broad range of studies that radiologists interpret in a community hospital setting appear to be far lower than earlier studies on selective radiographs indicated.
RATIONALE AND OBJECTIVES: Prospective studies of radiologists' interpretations of selected radiographs reported 20-40 years ago indicated error rates of 30% and higher. The authors retrospectively evaluated the interpretations of groups of radiologists and determined a range of rates of disagreement in interpretation. Quality assessment or recredentialing may add to the importance of such studies in the future. MATERIALS AND METHODS: Over a 7-year period, a team of radiologists reviewed imaging interpretations in the radiology departments of six community hospitals. Each review, which lasted about 3 days, included evaluation of the interpretations of a 3%-4% sample of the images read by the radiologists at these hospitals. Reading errors were quantitated and evaluated qualitatively. RESULTS: In a review of over 11,000 images read by 35 radiologists, the authors found a 4.4% mean rate of interpretation disagreement; only one radiologist had a mean rate above 8%. Qualitative analysis of the interpretation errors revealed a mean rate of 3.0% of errors that were considered to be below an acceptable standard of care. Radiologists whose errors included a relatively high proportion of false-positive findings tended to make relatively fewer total errors. CONCLUSION: Rates of disagreement for a broad range of studies that radiologists interpret in a community hospital setting appear to be far lower than earlier studies on selective radiographs indicated.