Observer agreement in the reporting of knee and lumbar spine magnetic resonance (MR) imaging examinations: Selectively trained MR radiographers and consultant radiologists compared with an index radiologist

Abstract

PurposeTo assess agreement between trained radiographers and consultant radiologists compared with an index radiologist when reporting on magnetic resonance imaging (MRI) examinations of the knee and lumbar spine and to examine the subsequent effect of discordant reports on patient management and outcome. MethodsAt York Hospital two MR radiographers, two consultant radiologists and an index radiologist reported on a prospective, random sample of 326 MRI examinations. The radiographers reported in clinical practice conditions and the radiologists during clinical practice. An independent consultant radiologist compared these reports with the index radiologist report for agreement. Orthopaedic surgeons then assessed whether the discordance between reports was clinically important. ResultsOverall observer agreement with the index radiologist was comparable between observers and ranged from 54% to 58%; for the knee it was 46–57% and for the lumbar spine was 56–66%. There was a very small observed difference of 0.6% (95% CI −11.9 to 13.0) in mean agreement between the radiographers and radiologists (P=0.860). For the knee, lumbar spine and overall, radiographers’ discordant reports, when compared with the index radiologist, were less likely to have a clinically important effect on patient outcome than the radiologists’ discordant reports. Less than 10% of observer's reports were sufficiently discordant with the index radiologist's reports to be clinically important. ConclusionCarefully selected MR radiographers with postgraduate education and training reported in clinical practice conditions on specific MRI examinations of the knee and lumbar spine to a level of agreement comparable with non-musculoskeletal consultant radiologists.