Degradation of gait autocorrelation as precocious marker of fall risk in Parkinson's disease

Abstract

Falls remain a major concern in Parkinson disease (PD). However, no precocious marker of fall risk has been highlighted. Recently, complex temporal organization of stride duration variability has been demonstrated, displaying long-range autocorrelations (LRA). The breakdown of such temporal dynamics is thought to be associated to dynamic instability in locomotion. However, no studies have included the analysis of stride duration variability in the functional assessment of patients presenting a neurological disease, such as PD. Temporal organization (LRA) and magnitude (CV and SD) of stride duration variability, collected from 20 parkinsonian patients walking on overground at a comfortable speed, were studied. The presence of LRA was based on scaling properties of the series variability (Hurst exponent) and the shape of the power spectral density (α exponent). Functional status, objective and subjective measures of balance were collected using MDS-UPDRS (and PIGD), modified H&Y scale, BESTest and ABC Scale, respectively. In order to precise the relationship between stride duration variability and the dynamic stability, a principal component analysis was applied. LRA were highlighted in all patients and excellent correlations were observed between LRA (factor loading: Hurst = 0,907; α = 0,931), H&Y scale (−0,914), BESTest score (0,867) and PIGD (−0,741), while no significant correlations were observed with age and gait speed. CV and SD stride duration were inversely correlated with ABC Scale. Supporting firmly the relationship with dynamic stability, strong correlations between LRA, disease severity and balance status were demonstrated for the first time in PD. Taken together with clinical tests, magnitude and organization of variability allow a more comprehensive description of specific deficits increasing fall risk than either method could provide alone. While magnitude of variability could be a marker of the degree of automaticity of gait, LRA could be considered as the reflection of dynamic stability. The quantification of stride duration variability can thus constitute an easy and effective tool to objectively and quantitatively provide a clinical measure of fall risk in PD.