Angular Measurement Error Due to Different Measuring Devices

Abstract

Angular measurements are commonly used in orthopaedic surgery. No study has addressed measurement variability due to the measurement device itself. It was the purpose of this study to assess measurement variability of articulated versus fixed devices. Three articulated and 4 fixed goniometers were randomly selected. Thirty-two different angles, ranging from a few degrees to nearly 180 degrees, were drawn with a standard soft-lead marking pencil. The angles were measured by 5 different observers with 7 different goniometers on 2 separate occasions separated by a minimum of 3 weeks. We wished to determine whether the variability of any goniometer was within a +/-1-degree range. There were 1023 (91.3%) absolute differences of 1 degree or less and 97 (8.7%) of more than 1 degree. Intraobserver agreement was 92.0%; 96.9% for fixed and 84.0% for hinged goniometers. Interobserver agreement was of 90.7%; 96.2% for the fixed and 83.4% for the hinged goniometers. Intragoniometer agreement was 91.3%; 96.9% for fixed and 84.0% for hinged goniometers. Intergoniometer agreement was 87.0% with fixed goniometers demonstrating better agreement than hinged goniometers. The overall intraobserver and interobserver measurement variability was +/-2.33 and +/-2.26 degrees, respectively; the overall intragoniometer and intergoniometer measurement variability was +/-2.26 and +/-2.30 degrees, respectively. For fixed goniometers, the intergoniometer measurement variability is +/-2.0 degrees, for hinged goniometers, +/-2.9 degrees, and when using both fixed and hinged goniometers, +/-2.4 degrees. Thus, the fixed goniometers are the ideal type with overall better agreement and measurement variability. Any one orthopaedic surgeon should use the same goniometer at all times, preferably a fixed type, so that measurement variability can be reduced by +/-2.0 degrees. If a physician uses a particular published measurement variability from the literature in which different goniometers were used, then the intraobserver measurement variability will be less than the published value by approximately +/-2 degrees. This is important when faced with the question of a change in an angular measurement being a true change or simply a reflection of measurement error.