Characteristics of objectively-measured naturalistic light exposure patterns in U.S. adults: A cross-sectional analysis of two cohorts

Abstract

Light is an environmental feature important for human physiology. Investigation of how light affects population health requires exposure assessment and personal biomonitoring efforts. Here, we derived measures of amount, duration, regularity, and timing from objective personal light (lux) measurement in >4000 participants across two United States (US)-based cohort studies, the Multi-Ethnic Study of Atherosclerosis (MESA) and the Hispanic Community Health Study / Study of Latinos (HCHS/SOL), encompassing eight geographic regions. Objective light and actigraphy data were collected over a week using wrist-worn devices (Actiwatch Spectrum). Cohort-stratified light exposure metrics were analyzed in relation to sex, season, time-of-day, location, and demographic and sleep health characteristics using Spearman correlation and linear and logistic regressions (separately by cohort) adjusted for age, sex (where applicable), and exam site. Light exposure showed sex-specific patterns and had seasonal, diurnal, geographic, and demographic and sleep health-related correlates. Results between independent cohorts were strongly consistent, supporting the utility and feasibility of light biomonitoring. These findings provide a fundamental first characterization of light exposure patterns in a large US sample and will inform future work to incorporate light as a biologically relevant exposure in environmental public health and key component of the human exposome.

Keywords: Chronobiology; Circadian; Illumination; Photoperiod; Seasonality; Sleep disorders.

Fig. 1.

Example of white light exposure for one participant across 6 days (12 AM–12 AM) and derived light variables. Black lines represent log10-transformed white light illuminance (lux+1) values. Horizontal blue bars represent illuminance thresholds (≥10, ≥100, and ≥ 1000 lx) for calculating time above light thresholds (TALT). The span of each horizontal blue bar represents the daily individual photoperiod (IP) at the specified illuminance threshold (IP₁₀, IP₁₀₀, IP₁₀₀₀), the downward salmon arrows and upward tan arrows represent the first (FTL) and last timing (LTL) of light exposure at that threshold, respectively. The brown curly brackets represent the data used for the timestamps in calculating the mean light timing revised (MLiTR) at the corresponding illuminance threshold (≥10, ≥100, and ≥ 1000 lx).

Fig. 2.

Plots showing differences in light exposure by sex across age in HCHS/SOL and MESA. Males (dark orange) had higher levels of both average illuminance calculated (A) without and (B) with prior log₁₀-transformation and (C) duration spent in bright light (≥1000 lx, TALT₁₀₀₀) compared to females (light orange) across different ages. The shaded bands indicate 95 % confidence intervals.

Fig. 3.

Average log10-lux illuminance values in (A) HCHS/SOL and (B) MESA and by time of day, grouped by season of measurement. Grey line represents average light exposure during the winter season (November – January), blue line represents the spring season (February–April), the dark orange line represents the summer season (May–July), and the light orange line represents the fall season (August–October).

Fig. 4.

Heatmap plotting the Spearman correlations (p < 0.05) between light variables in HCHS/SOL (upper right triangle, n = 1931 with complete data) and MESA (lower left triangle, n = 1693 with complete data); diagonal black line indicates cohort-stratified correlations are within, not between, cohorts). Variables with a “log” prefix indicate that lux values were log10-transformed prior to metric calculation. Correlations with p > 0.05 are left blank. Blue color indicates positive correlation and red color indicates negative correlation. Abbreviations: FTL = first timing of light; HCHS/SOL = Hispanic Community Health Study / Study of Latinos; IP = individual photoperiod; LEDS = light exposure during sleep; LTL = Last timing of light; maxtemp = maximum daily temperature (degrees Celsius); MESA = Multi-Ethnic Study of Atherosclerosis; MLiTR = mean light timing revised; SD = standard deviation; TALT = time above light threshold.

Fig. 5.

Heatmap plotting the Spearman correlations (p < 0.05) with average and SD of log10-transformed illuminance in 2-h time windows in HCHS/SOL (upper right triangle) and MESA (lower left triangle; diagonal black line indicates cohort-stratified correlations are within, not between, cohorts). Correlations with p > 0.05 are left blank. Blue color indicates positive correlation and red color indicates negative correlation.

Fig. 6.

Heatmap plotting the Spearman correlations (p < 0.05) between age, BMI, sleep-related and light variables in HCHS/SOL and MESA (cohort-stratified; correlations are within, not between, cohorts). Variables with a “log” prefix indicate that lux values were log10-transformed prior to metric calculation. Correlations with p > 0.05 are left blank. Blue color indicates positive correlation and red color indicates negative correlation. Abbreviations: BMI = body mass index; ESS = Epworth Sleepiness Scale; FTL = first timing of light; HCHS/SOL = Hispanic Community Health Study / Study of Latinos; IP = individual photoperiod; IS = interdaily stability; IV = intradaily variability; L5 = activity during the 5 h of lowest activity; L5 start = start time of the 5 hours of lowest activity; LEDS = light exposure during sleep; LTL = last timing of light; M10 = activity during the 10 h of highest activity; M10 start = start time of the 10 hours of highest activity; MESA = Multi-Ethnic Study of Atherosclerosis; MEQ = Morningness-Eveningness questionnaire; MLiTR = mean light timing revised; RA = relative amplitude; SD = standard deviation; TALT = time above light threshold; WHIIRS = Women’s Health Initiative Insomnia Rating Scale.