Nighttime light is a growing anthropogenic health threat, particularly in urban areas. Limited evidence suggests that exposure to outdoor artificial light at night (ALAN) may be associated with people's mental health by disrupting sleep-wake cycles.
We assessed 1) the association between ALAN exposure and adults’ symptoms of depression and anxiety, 2) whether the association was modified by sex, age, and income, and 3) the mediating role of sleep problems.
We obtained cross-sectional data from 4,068 adults from the five largest Bulgarian cities. Depression and anxiety symptoms were measured using the 4-item Patient Health Questionnaire (PHQ-4). Sleep problems were self-reported based on three items. Outdoor ALAN at residential addresses was assessed using annual radiance levels obtained from satellite imagery. Regression models were adjusted for person-level characteristics, green space, and nitrogen dioxide (NO2). We also assessed effect modification by sex, age, and income. Using mediation analyses, we tested sleep problems as a mediator of the ALAN-PHQ-4 association.
Greater ALAN exposure in the fully adjusted model was marginally associated with higher PHQ-4 scores. We observed no effect modification. The mediator, sleep problems, was also positively associated with ALAN. The mediation of sleep problems was significantly positive. While the direct association was null, the total ALAN association was marginally and positively associated with PHQ-4 scores.
Our findings suggest a positive association between outdoor nighttime light pollution and mental health. Poor sleep quality is a possible pathway relating ALAN exposure to mental health. Considering the increasing ubiquity and intensity of urban nighttime illumination, light pollution-reducing policies may provide significant health benefits for urban populations.
We anticipated that ALAN exposure might harm people's mental health either directly or by disrupting sleep quality. We identified sleep problems with survey items addressing the respondents' sleep patterns over the two weeks preceding (Dzhambov et al., 2024). Three items probed issues with falling asleep, waking up and remaining sleepless during the night, and feeling unrested the following morning. Responses were given on a four-point Likert scale (1 = never, 2 = several days, 3 = more than half the days, 4 = nearly every day). We summed the item scores to obtain an overall sleep problem score ranging from 0 to 9. Higher scores indicate more sleep problems. The internal item consistency was good (Cronbach's alpha = 0.856; 95% CI: 0.843; 0.867).
Space-borne assessment of outdoor nighttime light
We used satellite imagery to identify outdoor artificial nighttime luminance (Kocifaj et al., 2023). Due to ALAN uncertainties introduced by first-generation remote sensing platforms (i.e., the Defense Meteorological Satellite Program Operational Linescan System [DMSP-OLS]), we used globally calibrated data from the day/night band on the panchromatic Visible Infrared Imaging Radiometer Suite (VIIRS) sensor operating on the Suomi National Polar-orbiting Partnership satellite (Miller et al., 2012). We favored VIIRS data for its higher spatial resolution of ∼464 m compared to ∼1 km for DMSP-OLS (Levin et al., 2020), enhanced radiometric sensitivity, and in-flight sensor calibration (Elvidge et al., 2017; Levin et al., 2020).
To match our survey data as closely as possible while ensuring the exposure occurred before the survey took place and to mitigate seasonal variations, we used the twelve-month median radiance grid (in nW/cm2/sr) corrected for stray light. Google Earth Engine provided scenes from the 2022 ‘VCMSLCFG’ image collection sourced from the data archive hosted by the National Oceanic and Atmospheric Administration.
We estimated exposure to ALAN at respondents’ residential addresses. We geocoded the address locations using global positioning system-based tracking devices. The spatial accuracy of the address coordinates was evaluated based on auxiliary data (e.g., cadaster) and, if needed, manually corrected on a case-by-case basis. Since it is possible for a respondent to reside at the boundary of a cell and be more affected by an adjacent cell, we resampled the grid to 10 m using bilinear interpolation to match the resolution of the co-exposures (see below). Given the original ALAN data resolution, we used 500 m circular buffers centered on address locations as our main contextual unit to delineate environmental exposures. Our buffer size was comparable to a buffer size used elsewhere.
Comments
Post a Comment