Personal light exposure patterns and incidence of type 2 diabetes: analysis of 13 million hours of light sensor data and 670,000 person-years of prospective observation

Abstract

Background

Light at night disrupts circadian rhythms, and circadian disruption is a risk factor for type 2 diabetes. Whether personal light exposure predicts diabetes risk has not been demonstrated in a large prospective cohort. We therefore assessed whether personal light exposure patterns predicted risk of incident type 2 diabetes in UK Biobank participants, using ∼13 million hours of light sensor data.

Methods

Participants (N = 84,790, age (M ± SD) = 62.3 ± 7.9 years, 58% female) wore light sensors for one week, recording day and night light exposure. Circadian amplitude and phase were modeled from weekly light data. Incident type 2 diabetes was recorded (1997 cases; 7.9 ± 1.2 years follow-up; excluding diabetes cases prior to light-tracking). Risk of incident type 2 diabetes was assessed as a function of day and night light, circadian phase, and circadian amplitude, adjusting for age, sex, ethnicity, socioeconomic and lifestyle factors, and polygenic risk.

Findings

Compared to people with dark nights (0–50th percentiles), diabetes risk was incrementally higher across brighter night light exposure percentiles (50–70th: multivariable-adjusted HR = 1.29 [1.14–1.46]; 70–90th: 1.39 [1.24–1.57]; and 90–100th: 1.53 [1.32–1.77]). Diabetes risk was higher in people with lower modeled circadian amplitude (aHR = 1.07 [1.03–1.10] per SD), and with early or late circadian phase (aHR range: 1.06–1.26). Night light and polygenic risk independently predicted higher diabetes risk. The difference in diabetes risk between people with bright and dark nights was similar to the difference between people with low and moderate genetic risk.

Interpretation

Type 2 diabetes risk was higher in people exposed to brighter night light, and in people exposed to light patterns that may disrupt circadian rhythms. Avoidance of light at night could be a simple and cost-effective recommendation that mitigates risk of diabetes, even in those with high genetic risk.

Funding

Australian Government Research Training Program.