VALIDATION OF TECHNIQUES IN THE MEASUREMENT OF RANGE OF MOTION AFTER TOTAL KNEE ARTHROPLASTY: ROUGH ESTIMATE OR RELIABLE OUTCOME MEASURE?

Abstract

Introduction: Range of motion is an invariable outcome -measure in studies on total knee arthroplasty (TKA) and other knee surgery. Concluding that a certain change in motion equals a corresponding change in outcome may be invalid if true accuracy of current measuring techniques is unknown. This is integral to many studies. Surprisingly little has been done to validate these techniques.

Methods: Maximum extension and flexion were measured in 32 TKAs by four independent observers using three common techniques: visual estimate (VE), pocket and universal goniometers (PG and UG). Lateral radiographs in reproducible positions were measured using computer analysis, providing a gold standard for comparison with clinical measurements. The correlation coefficients and coefficients of reliability were calculated.

Results: There were no significant differences between observers using any method. Significant differences were found between each technique and radiographic measure (paired t-test, p< 0.001). Correlation coefficients were lower for extension estimates (0.76–0.80) than flexion (0.91–0.96). Coefficients of repeatability varied from 11.6 degrees to 12.1 degrees for extension measurements and from 13.8 degrees to 19.2 degrees for flexion measurements, with UG being the most accurate. The VE accuracy approached that of UG only at easily visualised angles such as 90 degrees. The coefficient of repeatability for radiographic measure was significantly lower at 2.9 degrees.

Conclusions: Clinical measurements of range of motion vary significantly from radiographic measurement, with the computer assisted radiographic measurement providing high reliability as the gold standard. UG is most accurate, followed by PG and VE. However, coefficients of repeatability were surprisingly large, indicating the degree of accuracy of each measurement technique and the necessary magnitude of difference for this to be outside measurement error. This has relevance for all outcome studies and everyday clinical practice.