Accuracy and inter-observer reliability of visual estimation compared to clinical goniometry of the elbow

To test the hypothesis that visual estimation by a trained observer is as accurate and reliable as clinical goniometry for measuring elbow range of motion. Instrument validity and inter-observer reliability of visual estimation was evaluated on a consecutive series of 50 elbow contractures. Four observers with different levels of elbow experience first estimated extension and flexion of the contracted elbows and then measured them with a blinded goniometer. Instrument validity for visually-based goniometry was extremely high. ICC scores were 0.97 for both extension and flexion estimations. Systematic error was negligible (1A degrees) with upper limits of agreement being 9A degrees (95% CI: 7A degrees-11A degrees) and 8A degrees (95% CI: 6A degrees-10A degrees), respectively, for extension and flexion. For the expert surgeon, 92% of the visual estimates were within 5A degrees of the value obtained by clinical goniometry. Between experienced observers (elbow surgeon and physician assistant), the ICC's were very high-0.96 for extension and 0.93 for flexion. The systematic errors were low, from -1A degrees to 1A degrees with upper limit of agreement being 11A degrees (95% CI: 8A degrees aEuro14A degrees). However, agreement was poor between an inexperienced study coordinator and the others (ICC's: 0.51-0.38, systematic errors: 8A degrees-18A degrees, upper limit of agreement: 32A degrees-40A degrees). The accuracy of the visual estimations made by the experienced elbow surgeon was as good as the measurements taken with a goniometer by the physician assistant or the clinical fellow and better than those taken by an inexperienced study coordinator. The trained human eye is highly capable of accurately estimating the range of motion of the elbow, compared to conventional clinical goniometry, depending on the experience of the observer. Diagnostic study, Level II.