The Optimal Measurement Period of Actigraphy for Circadian Rhythm in Relation to Adiposity: A Retrospective Case-Control Study

Hai-Hua Chuang,Yu-Hsuan Lin,Li-Ang Lee,Hsiang-Chih Chang,Guan-Jie She,Chen Lin

DOI: https://doi.org/10.1016/j.sleep.2024.07.025

IF: 4.842

2024-07-26

Sleep Medicine

Abstract:Background This study focused on the relationship between adiposity and Rest-Activity Rhythms (RAR), utilizing both parametric cosine-based models and non-parametric algorithms. The emphasis was on the impact of varying measurement periods (7 to 28 days) on this relationship. Methods We retrieved actigraphy data from two datasets, encompassing a diverse cohort recruited from an obesity outpatient clinic and a workplace health promotion program. Participants were required to wear a research-grade wrist actigraphy device continuously for a minimum of four weeks. The final dataset included 115 individuals (mean age 40.7 ± 9.5 years, 51% female). We employed both parametric and non-parametric methods to quantify RAR using six standard variables. Additionally, the study evaluated the correlations between three key adiposity indices — Body Mass Index (BMI), Visceral Adipose Tissue (VAT) area, and Body Fat Percentage (BF%) — and circadian rhythm indicators, controlling for factors like physical activity, age, and gender. Results The obesity group displayed a significantly lower relative amplitude (RA) as per non-parametric algorithm findings, with a decreased amplitude noted in the parametric algorithm analysis, in comparison to the overweight and control groups. The relationship between circadian rhythm indicators and adiposity metrics over 7- to 28-day periods was examined. A notable negative correlation was observed between RA and both BMI and VAT, while correlation coefficients between adiposity indicators and non-parametric circadian parameters increased with extended durations of actigraphy data. Specifically, RA over a 28-day period was significantly correlated with BF%, a trend not seen in the 7-day measurement (p = 0.094) in multivariate linear regression. The strength of the correlation between BF% and 28-day RA was more pronounced than that in the 7-day period (p = 0.044). However, replacing RA with amplitude as per parametric cosinor fitting yielded no significant correlations for any of the measurement periods. Conclusion The study concludes that a 28-day measurement period more effectively captures the link between disrupted circadian rhythms and adiposity. Non-parametric algorithms, in particular, were more effective in characterizing disrupted circadian rhythms, especially when extending the measurement period beyond the standard 7 days.

clinical neurology