How many gait initiation trials are necessary to reliably detect anticipatory postural adjustments and first step characteristics in healthy elderly and people with Parkinson's disease?

Background: The gait initiation (GI) process can be characterized by anticipatory postural adjustments (APAs) and first step characteristics. However, even within a constrained environment, it is unclear how many trials are necessary to obtain a reliable measurement of the GI process within one assessment.

Research Question: How many gait initiation trials are necessary to reliably detect APAs and first step characteristics in healthy elderly (HC) and people with Parkinson's disease with Freezing of Gait (PD + FOG) under single (ST) and dual task (DT) conditions and are there any potential systematic errors?

Methods: Thirty-eight PD + FOG (ON-medication) and 30 HC performed 5 trials of GI under ST and DT (auditory stroop test). APAs and first-step-outcomes were captured with IMUs placed on the lower back and on each foot. Intraclass correlation coefficients (ICCs) and the standard error of measurement (SEM) were computed to investigate reliability and mixed model analysis to find potential systematic errors. Additionally, we computed an estimation for the number of necessary trials to reach acceptable reliability (ICC = 0.75) for each outcome.

Results: ICCs varied from low reliability to excellent reliability across outcomes in PD + FOG and HC. ICCs were comparable under ST and DT for most outcomes. SEM results confirmed the ICC results. A systematic error was found for the first trial in first step ROM. Number of necessary trials varied largely across outcomes.

Significance: Within-session reliability varied across outcomes but was similar for PD + FOG and HC, and ST and DT. ML size of APA and first step ROM were most reliable, whereas APA duration and latency were least reliable. Depending on the outcome of interest, future studies should conduct multiple trials of GI to increase reliability.