Ultrasonographic examination of acute soft tissue lesions in the elbow has good inter-rater reliability and acceptable agreement with magnetic resonance imaging

Abstract

BackgroundUltrasound (US) has been suggested a valuable complement to clinical and radiological examinations in elbow trauma. Magnetic resonance imaging (MRI) has been the method of choice, despite fair to moderate interrater reliability (IRR). US has potential advantages but is assessor dependent and IRR scarcely examined. The primary aim of the present study was to investigate IRR for US and secondarily inter-observer agreement (IOA) between US and MRI in the acute phase after elbow trauma. Acute phase was defined as 2 weeks and, if applicable, the following weekend. The hypothesis was that US reliability would be at least substantial for complete muscle or ligament lesions. Methods116 patients (50 men, median age 47 [range 19-87] years) suffering an elbow trauma with dislocation and/or fracture were included. Exclusion criteria were prior injury to the same elbow, and US and/or MRI not possible within 16 days. During US the condition of muscle origins at the epicondyles and collateral and annular ligament complexes was recorded in a pre-designed protocol, with the alternatives intact, partially or completely torn. 72 patients had a second US examination the same day by an independent upper extremity surgeon. 58 of the 116 patients underwent an MRI before or after the US, evaluated by 2 radiologists using the same protocol. IOA and IRR between assessors and modalities were analyzed with kappa statistics and interpreted according to Landis and Koch. Perfect agreement (PA) was reported in percent. ResultsUS examination within 2 weeks was feasible with tolerable discomfort. Defining muscle origins and ligaments as intact or completely torn the US IRR ranged from substantial to near perfect (kappa 0,63-1; PA 93-100%). Intact tissues vs tear (partial and complete tear combined) or intact vs partial vs complete tear resulted in kappa values from moderate to substantial and PA 74%-96% with lowest reliability for the muscle origins. The IOA between MRI and US ranged from fair to near perfect for no tear vs complete tear (kappa 0,25-1; PA 65-100%). Agreement between no tear and tear (partial and complete together) ranged from fair to substantial (0,25-0,66; PA 63-89%) and no tear vs partial or complete tear ranged from fair to moderate (0,25-0,53; PA 50-79%). ConclusionUS in the acute setting is suitable and reliable for diagnosis of ligament injuries in the elbow and is in addition fast, cheap and easily accessible. The agreement with MRI seems to vary with the structure assessed and severity of the lesions, ranging from fair to near perfect.