Test-retest reliability of trunk accelerometric gait analysis.

The purpose of this study was to determine the test-retest reliability of a trunk accelerometric gait analysis in healthy subjects. Accelerations were measured during walking using a triaxial accelerometer mounted on the lumbar spine of the subjects. Six men and 14 women (mean age 35.2; range 18-57) underwent the same protocol on 2 consecutive days. The raw acceleration signals from six self-selected walking speeds were transformed into a horizontal-vertical coordinate system to remove unwanted variability caused by gravity. Acceleration root mean square values, cadences, step and stride lengths were then computed and interpolated using quadratic curve fits and point estimates were calculated at a standardised walking speed of 1.35 m/s. Relative reliability was determined using two models of intraclass correlation coefficients (ICC(1,1) and ICC(3,1)) to assess any systematic shifts and absolute reliability was determined using measurement error (ME). The results of the study showed high ICC values (0.77-0.96) and ME values of 0.007-0.01 g for mean acceleration; 0.009 m for step lengths; 0.022 m for stride length and 1.644 step/min for cadences. In conclusion, the method was found to be reliable and may have a definite potential in clinical gait analysis. Copyright 2003 Elsevier B.V.