Over the past few decades, rapid urbanization and industrialization have significantly increased artificial light at night (ALAN) in global cities (Cauwels et al., 2014). The introduction of artificial lighting has altered the spectrum, intensity, and timing of light. Previous studies reported that nearly 80 % of the U.S. population lives in areas where the natural night sky is no longer visible, and about 40 % reside in regions where artificial light hinders proper nocturnal adaptation of the human eyes (Falchi et al., 2016). Furthermore, outdoor ALAN continues to expand, with some cities experiencing annual increases in light pollution ranging from 5 % to 20 % during the latter half of the 20th century (Hölker et al., 2010). ALAN has become a ubiquitous environmental pollutant in modern society.

Unlike other pollutants, ALAN does not pose direct toxic effects. However, emerging evidence suggests that ALAN exerts multifaceted impacts on human health through complex pathways. A recent systematic evidence map categorizes these impacts into three domains: (1) direct physiological effects (e.g., circadian disruption, melatonin suppression), (2) psychological consequences (e.g., depression, anxiety), and (3) social determinants of health, including neighborhood safety and environmental justice (Liu et al., 2023). Additionally, using ALAN in certain ways may also lead to opposite health effects. For example, short exposure to ALAN may help phase adaptation (Chinoy et al., 2016) and reduce work-related risks (Figueiro et al., 2016). The complexity and diversity of these issues have kept growing with more studies in the last two decades. Among these, the role of ALAN in mental health is a domain accounting for 23.9 % of recent ALAN-health research (Liu et al., 2023).

Indoor ALAN has been shown to acutely disrupt sleep by suppressing melatonin production and delaying circadian rhythms (Obayashi et al., 2014). The psychological consequences due to ALAN are another area of interest. Melatonin suppression caused by ALAN exposure is believed to disturb circadian and endocrine hormonal balance, negatively affecting mental health (Nelson and Chbeir, 2018). Additionally, light exposure can directly influence mood, independent of circadian and sleep changes (Fernandez et al., 2018). For instance, it can affect mood by regulating the secretion of neurotransmitters like serotonin (Blume et al., 2019). Additionally, ALAN exposure has also been associated with the dysregulation of the immune system, which can modulate brain functions, including cognition and emotion (Navara and Nelson, 2007). Previous studies have suggested that ALAN can significantly impact public health, with effects becoming more pronounced as the extent and intensity of artificial lighting increase (Kyba et al., 2017). It is postulated that the adverse effects of ALAN are not confined to indoor lighting; outdoor ALAN (O-ALAN) in urban environments may also affect human health. Although the intensity of O-ALAN is generally lower, even dim light comparable to candlelight can influence the human circadian rhythm, highlighting its potential to affect human health (Duffy and Wright, 2005). O-ALAN can also impact indoor environments when curtains or shades fail to block outdoor light effectively (Stevens, 2011). Additionally, O-ALAN can expose those who are outdoors late at night to higher light levels. Since the near-inevitability of exposure to O-ALAN affects almost everyone in illuminated areas, it poses great concerns for public health, which, however, has not been well addressed yet. Thus, the major objective of this study is to evaluate O-ALAN as a form of pollution to investigate its association and potential causative relationships with human health, particularly in terms of sleep and mental health.

The existing cross-sectional epidemiological studies on the relationship between O-ALAN and sleep consistently provided evidence of O-ALAN's detrimental impact on sleep. For instance, a study found that individuals living in areas with high O-ALAN levels were more likely to report shorter sleep durations in the US (Xiao et al., 2020). Research also determined a nonlinear relationship between O-ALAN exposure and sleep parameters such as duration and quality (Ohayon and Milesi, 2016; Paksarian et al., 2020), and factors like low income and noise may exacerbate O-ALAN's impact on sleep (Gabinet and Portnov, 2021). Additionally, a study from South Korea found a statistically significant association between O-ALAN exposure and the use of sleep-inducing medications (J.-Y. Min and Min, 2018). The overall body of evidence suggests that O-ALAN could contribute to sleep deprivation. Even few epidemiological studies have examined the relationship between O-ALAN and mental health, and the findings remain inconsistent (Wang et al., 2023). A study from South Korea found that adults living in areas with higher levels of light pollution were more likely to experience depressive symptoms (J. Min and Min, 2018). Similarly, a positive correlation between O-ALAN exposure and feelings of low mood, lack of enthusiasm, and fatigue was revealed in a sample from the United Kingdom (Liao et al., 2023). In contrast, a study of the Dutch population found that while depressive symptoms increased with rising levels of O-ALAN, the association disappeared after adjusting for nitrogen dioxide (NO2) (Helbich et al., 2020). Another study indicated a weak link between O-ALAN exposure and poor mental health in rural U.S. areas affected by shale gas development (Boslett et al., 2021). Given the current inconclusive evidence, it is imperative to identify possible relationships between O-ALAN and mental health (Bozejko et al., 2023).

Most of these studies have been conducted on small-scale and relatively homogeneous populations or under controlled laboratory conditions. Consequently, the generalizability of the observed relationships and their applicability to real-world settings remain uncertain. Given the diversity in human lifestyles and behavioral patterns shaped by geographic, climatic, cultural, and social contexts, a comprehensive understanding of how O-ALAN affects human health requires evidence from both medical and social sciences. Research employing panel data and methods better suited for testing causal relationships is particularly needed (Wang et al., 2023). Ecological studies, in particular, can capture populations that cohort studies may not fully represent. In this context, ecological research may offer stronger epidemiological evidence than small-sample cohort studies (Bozejko et al., 2023). Existing ecological studies use average pixel values of nighttime light in an administrative unit as a proxy for O-ALAN level (Paksarian et al., 2020; Patel, 2019). However, significant variation in light intensity and population density within these units can lead to considerable misestimation of actual population exposure. Additionally, some studies have found different associations between O-ALAN and sleep deprivation in areas with varying income levels (Xiao et al., 2020), yet no comparable analyses have been conducted across regions with differing educational attainment. Furthermore, endogeneity may arise from unmeasured factors that simultaneously influence both O-ALAN and health outcomes. For example, socio-economic conditions could drive both light pollution and other determinants of sleep and mental health. To address the endogeneity stemming from such omitted variables, it is appropriate to identify instrumental variables that directly associate and even influence nighttime light, which may further affect human health.

Particularly, in this study, we aim to quantify the potential effects of O-ALAN, as a form of pollution, on sleep and mental health at the census tract level. To achieve that, we calculated population-weighted O-ALAN exposure for census tracts in the 500 largest cities in the contiguous US from 2013 to 2019. To account for endogeneity related to socio-economic conditions, we employed a Two-Stage Least Squares (2SLS) instrumental variable approach to examine the relationship between regional O-ALAN exposure and the prevalence of insufficient sleep and poor mental health. The major contributions of this study include: (i) determining whether O-ALAN at the census tract level is associated with insufficient sleep and poor mental health; and (ii) assessing whether the relationship between O-ALAN and these health outcomes varies according to other social or environmental factors, such as differences in educational attainment and urbanization levels.