Test-retest reliability of knee kinesthesia in healthy adults

Abstract

Background: Sensory information from mechanoreceptors in the skin, muscles, tendons, and joint structures plays an important role in joint stability. A joint injury can lead to disruption of the sensory system, which can be measured by proprioceptive acuity. When evaluating proprioception, assessment tools need to be reliable. The aim of this study was to assess the test-retest reliability of a device designed to measure knee proprioception.

Methods: Twenty-four uninjured individuals (14 women and 10 men) were examined with regard to test-retest reliability of knee kinesthesia, measured by the threshold to detection of passive motion (TDPM). Measurements were performed towards extension and flexion from the two starting positions, 20 degrees and 40 degrees knee joint flexion, giving four variables. The mean difference between test and retest together with the 95% confidence interval (test 2 minus test 1), the intraclass correlation coefficient (ICC2,1), and Bland and Altman graphs with limits of agreement, were used as statistical methods for assessing test-retest reliability.

Results: The intraclass correlation coefficients ranged from 0.59 to 0.70 in all variables except one. No difference was found between test and retest in three of the four TDPM variables. TDPM would need to decrease between 10% and 38%, and increase between 17% and 24% in groups of uninjured subjects to be 95% confident of detecting a real change. The limits of agreement were rather wide in all variables. The variables associated with the 20-degree starting position tended to have higher intraclass correlation coefficients and narrower limits of agreement than those associated with 40 degrees.

Conclusion: Three TDPM variables were considered reliable for observing change in groups of subjects without pathology. However, the limits of agreement revealed that small changes in an individual's performance cannot be detected. The higher intraclass correlation coefficients and the narrower limits of agreement in the variables associated with the starting position of 20 degrees knee joint flexion, indicate that these variables are more reliable than those associated with 40 degrees. We, therefore, recommend that the TDPM be measured with a 20-degree starting position.

Figure 1

Knee kinesthesia tested in a lateral decubitus position on a specially designed platform. The subject is a model who did not participate in the study.

Figure 2

Bland & Altman graph with limits of agreement (LOA). The differences between test sessions 1 and 2 (test 2 minus test 1) plotted against their mean for each subject for TE20 (degrees) in 24 uninjured subjects, together with the 95% confidence interval (CI) and the 95% LOA. In this figure, the differences are generally increasing with their means (heteroscedasticity). Note: several subjects have the same value.

Figure 3

Bland & Altman graph with limits of agreement (LOA) after log transformation. The differences between test sessions 1 and 2 (test 2 minus test 1) plotted against their mean for each subject for TE20 after log_e transformation with the 95% LOA. Note: several subjects have the same value.