Exploration of the test-retest reliability and agreement between the Microsoft Kinect and vertically orientated Perometer when measuring lower limb volume in a healthy student population.

Abstract

Question: Does the Microsoft Kinect exhibit high levels of test-retest reliability and agreement when compared to the vertically orientated Perometer in a healthy population? Design: Between-methods agreement and test-retest reliability study Setting: University setting Participants: Fifteen healthy University students were recruited using convenience sampling with no drop outs. Interventions: All participants' dominant lower limb volume was measured using the vertically orientated Perometer and the Microsoft Kinect. Outcome measures: The test-retest reliability of the two measurements was determined using the ICC (3,1). The measurement error was determined using the standard error of measurement (SEM). The Bland-Altman difference plot was used to determine bias between the two methods and its 95% confidence interval (CI). Results: The test-retest reliability for the Perometer and the Microsoft Kinect was ICC (3,1) = 0.994 and ICC (3,1) = 0.956 respectively. The measurement error for the Perometer and Microsoft Kinect were clinically acceptable (SEM = 146ml and 418ml respectively). There was a lack of agreement between the Perometer and the Microsoft Kinect. The Bland-Altman difference plot showed the Microsoft Kinect overestimated lower limb volume by 626ml compared to the Perometer. A 95% CI of -827.8ml to 2,079ml was observed. Conclusion: There was a lack of agreement between the vertically orientated Perometer and the Microsoft Kinect concluding therefore that they should not be used interchangeably. Both measurement tools showed high levels of test-retest reliability.