Non-parametric actigraphy-derived measures differ in dementia with Lewy bodies compared to Alzheimer’s dementia: a feasibility study

G.J. Elder, N. Santhi, K. Olsen, D. Polasek, S. Doyle, J.-P. Taylor

Research output: Contribution to conferenceAbstractpeer-review


Introduction: Dementia with Lewy bodies (DLB) is the second most common form of dementia. Sleep disturbances are highly prevalent in DLB and are more severe relative to other neurodegenerative dementias including Alzheimer’s dementia (AD).

DLB studies typically use subjective informant measures to assess sleep. Where objective sleep has been assessed, this has normally been done within a sleep laboratory or hospital environment on a single night, which can be disruptive, unrepresentative of habitual sleep and cause patient and caregiver burden.

One possibility is the use of actigraphy. This is a validated marker of the sleep-wake cycle and is a feasible method of assessing sleep in dementia populations, including DLB. To date, no studies have specifically examined non-parametric methods of actigraphy analysis in DLB. These can provide a comprehensive description of the circadian rest-activity pattern. This is important as non-parametric measures may be more sensitive to change than standard actigraphic measures of sleep continuity, and therefore have utility as a diagnostic/prognostic tool, or clinical trial outcome measure.

The aim of this study was to: 1) assess the feasibility of deriving non-parametric measures from DLB actigraphy data, and 2) compare these to an AD group. It was hypothesised that non-parametric measures would differ between DLB and AD.

Methods: One week of actigraphy data was continuously collected from mild-to-moderate probable DLB participants (n = 8) and probable AD participants (n = 6), using wrist-worn accelerometers (GENEActiv, Activinsights, UK)

Non-parametric measures were derived from actigraphy data using the nparACT package (v0.8; Blume et al., 2016) for R (R Core Team, 2021). Specifically, interdaily stability (IS), intradaily variability (IV), relative amplitude of activity (RA), 5 hours with the lowest actigraphy amplitude (L5) and 10 hours with the highest average amplitude (M10) values were calculated. These were compared between DLB and AD groups using non-parametric Mann-Whitney U-tests.

Results: All participants tolerated wearing accelerometers and it was feasible to obtain non-parametric rest-activity values from actigraphy data. Relative to AD, RA values were significantly lower (DLB: M = 0.34, SD = 0.25; AD: M = 0.69, SD = 0.14; p = .01) and M10 values were also lower in DLB (DLB: M = 32.72, SD = 24.40, AD: M = 89.29, SD = 22.31; p < .01). There were no significant between-group differences in IS, IV or L5 values.

Conclusions: It is feasible to derive non-parametric rest-activity measures from actigraphic data collected from individuals with DLB. The relative amplitude of activity, and 10 hours with the highest average amplitude were significantly lower in DLB compared to AD. This suggests that specific circadian markers derived from actigraphy may be differentially affected in DLB. This is clinically relevant and further work should examine if these measures are potentially a suitable DLB prognostic or diagnostic target, or if they represent a sensitive outcome measure in future clinical trials.

Acknowledgements: This study was funded by Alzheimer’s Research UK, Alzheimer’s Research UK North Network Centre, British Sleep Society, Northumbria University Graduate Futures, and the NIHR Newcastle Biomedical Research Centre.
Original languageEnglish
Number of pages1
Publication statusPublished - 1 Dec 2022

Cite this