Trajectories of Sleep Duration, Sleep Onset Timing, and Continuous Glucose Monitoring in Adults

Abstract

Importance: Understanding the interplay between trajectories of sleep duration, sleep onset timing, and glycemic dynamics is crucial for improving preventive strategies against diabetes and related metabolic diseases.

Objective: To examine the associations of sleep duration and onset timing trajectories with continuous glucose monitoring (CGM)-derived glycemic metrics in adults.

Design, setting, and participants: This cohort study analyzed data collected from January 2014 to December 2023 in the Guangzhou Nutrition and Health Study, a prospective cohort in Guangdong province, China, among participants aged 46 to 83. Participants who had repeated sleep assessments at several study visits and were equipped with CGM devices at the last visit were included. Data analyses were conducted between January and June 2024.

Exposures: The trajectories of sleep duration and onset timing were constructed using self-report sleep duration and sleep onset timing, recorded at multiple study visit points.

Main outcomes and measures: Measurements of glycemic variability and glycemic control were collected using a masked CGM device worn by patients for 14 consecutive days. Huber robust regression models were used to assess the associations between sleep trajectories and CGM-derived metrics.

Results: In this study of 1156 participants (mean [SD] age, 63.0 [5.1] years, 816 [70.6%] women), we identified 4 distinct sleep duration trajectory groups: severe inadequate, moderate inadequate, mild inadequate, and adequate. Severe sleep inadequacy was associated with an increment of glycemic variability indicators: 2.87% (95% CI, 1.23%-4.50%) for coefficient of variation and 0.06 (95% CI, 0.02-0.09) mmol/L for mean of daily differences. We found 2 trajectories of sleep onset timing: persistent early and persistent late groups. Late sleep onset was associated with larger coefficient of variation (β = 1.18%; 95% CI, 0.36%-2.01%) and mean of daily differences (β = 0.02 mmol/L; 95% CI, 0.01-0.04 mmol/L). Inappropriate sleep duration and timing trajectories in combination were associated with greater glycemic variability.

Conclusions and relevance: In this cohort study of middle-aged and older participants, persistent inadequate sleep duration and late sleep onset, whether alone or in combination, were associated with greater glycemic variability. These findings emphasize the importance of considering both sleep duration and timing for optimizing glycemic control in the general population.

Figure 1.. Description of Sleep Duration Trajectories at Baseline, First Follow-Up, and Second Follow-Up

The latent class growth modeling, a group-based modeling approach (SAS Proc Traj), was used to identify subgroups that shared a similar sleep duration trajectory from baseline to follow-up visits. The model with 4 trajectories showed the best fit to the data, including a severe inadequate group, a moderate inadequate group, a mild inadequate group, and an adequate group.

Figure 2.. Association of the Sleep Duration Trajectories With Continuous Glucose Monitoring–Derived Metrics

Association of long-term sleep duration patterns with glycemic variability, as qualified by coefficient of variation, SD, mean amplitude of glucose excursions, and mean of daily differences. β coefficients (95% CI) were derived from Huber robust regression models for sleep duration trajectories, severe inadequate sleep, moderate inadequate sleep, and mild inadequate sleep vs adequate sleep. Covariates included age, sex, body mass index, total energy intake, physical activity, income, education, smoking, alcohol drinking, and tea and coffee consumption. Multiple comparisons were controlled by false discovery rate of 0.05. Coefficient of variation was measured in percentage and SD, mean amplit ude, and mean of daily differences were measured in mmol/L. SI conversion factor: To convert glucose to mg/dL, divide by 0.0555. ^aFalse discovery rate < 0.05.

Figure 3.. Joint Associations of Long-Term Sleep Duration Trajectories and Sleep Onset Timing Trajectories With Glycemic Variability

Joint association of sleep patterns with glycemic variability, as qualified by CV, SD, MAGE, and MODD. Covariates included age, sex, body mass index, total energy intake, physical activity, income, education, smoking, alcohol drinking, and tea and coffee consumption. SI conversion factor: To convert glucose to mg/dL, divide by 0.0555. CV indicates coefficient of variation; MAGE, mean amplitude of glucose excursions; MODD, mean of daily differences.