Association between circadian physical activity trajectories and incident type 2 diabetes in the UK Biobank

Abstract

Physical activity (PA) is linked to a decreased risk of type 2 diabetes mellitus (T2DM). However, the influence of circadian PA trajectories remains uncertain. This study aims to explore the optimal circadian PA trajectory pattern for reducing the risk of T2DM. Methods: A total of 502,400 participants were recruited from the UK Biobank between 2006 and 2010, and 102,323 participants provided valid accelerometer-captured acceleration data. After excluding individuals with prior T2DM, 99,532 participants were included in the final analysis. We initially investigated the association between PA intensity at 24 hourly time points and T2DM. Subsequently, PA trajectories were identified using K-means cluster analysis. Cox proportional hazard models were employed to estimate hazard ratios (HR). Four distinct PA trajectories were identified: consistently low, single peak, double peak, and intense trajectories. Compared to consistently low, single peak, double peak and intense PA trajectory reduced the risk of T2DM progressively. Sensitivity analyses, further excluding individuals with glycated hemoglobin (HbA1c) ≥ 6.5% or random glucose ≥ 11.1 mmol/L and adjusted for daily average acceleration, yielded consistent results. This confirms that the ideal circadian PA trajectory serves as a protective factor, independently of PA intensity. Subgroup analyses indicated that these effects were more pronounced in men and individuals with eGFR < 60 mL/(min*1.73 m²). In conclusion, ideal circadian PA trajectory patterns (especially intense and then double peak) reduced risk of T2DM.

Figure 1

Trajectory groups of PA patterns in this study. Trajectory Cluster were based on continuous monitoring acceleration data and circadian rhythm temporal distribution from accelerometer over a span of 24 h.

Figure 2

Hourly PA and T2DM risk distribution in 24 h. (A) shows the risk on incident T2DM risk in mod3; (B) also shows association of T2DM risk with 24-hourly acceleration in mod1 and mod2. Model 1 adjusted for gender, age, and ethnicity. Model 2 further adjusted for education level, income, BMI, and smoking based on Model 1. Model 3 further adjusted for eGFR, CRP, and medical history of cardiovascular disease and cancer based on Model 2.

Figure 3

Association of incident T2DM with PA trajectories in mod1, mod2 and mod3. Model 1 adjusted for gender, age, and ethnicity. Model 2 further adjusted for education level, income, BMI, and smoking based on Model 1. Model 3 further adjusted for eGFR, CRP, and medical history of cardiovascular disease and cancer based on Model 2.