Less Is More - Estimation of the Number of Strides Required to Assess Gait Variability in Spatially Confined Settings

Background: Gait variability is a key marker of gait function and can be assessed using sensor-based methods. In clinical settings, limited space and patient condition often restrict walking distance, leading to the use of turning paradigms. However, many gait analysis systems exclude turns algorithmically. This study examined how turns affect sensor-based measures of gait variability.

Methods: Continuous inertial sensor recordings were obtained from 31 patients with movement disorders and 162 healthy elderly individuals during level walking with 180° turns. Turn-detection accuracy was verified, and strides before and after turns were compared with overall stride averages. The coefficient of variation (CoV) of stride length and stride time was calculated for all strides, between-turn segments, and cumulatively. Variance and consistency analyses were used to estimate the number of strides required for reliable variability measures.

Results: Undetected turns in a small subset of participants led to artificially increased CoV values. Even when turns were excluded, strides immediately before and after turns were shorter and longer in duration, respectively, increasing CoV and confidence intervals. A more generous turn-exclusion approach reduced these effects. Reliable CoV estimates were achieved after approximately 20 strides in all groups, and fewer in patients when using extended turn exclusion.

Conclusion: Including turns to extend walking distance does not necessarily improve the reliability of gait variability measures. Perturbations around turns may inflate variability, particularly in pathological gait. Short walking bouts of approximately 15 strides appear sufficient for reliable gait variability assessment.