Outdoor light at night and depressive symptoms in male veterans: a multicenter cross-sectional study in China

ABSTRACT

Previous studies have indicated depression was associated with environmental exposures, but evidence is limited for the association between outdoor light at night (LAN) and depression. This study aims to examine the association between long-term outdoor LAN exposure and depressive symptoms using data from the Chinese Veteran Clinical Research platform. A total of 6445 male veterans were selected from 277 veteran communities in 18 cities of China during 2009‒2011. Depressive symptoms were evaluated using the Chinese version of the Center for Epidemiological Studies Depression scale. Outdoor LAN was estimated using the Global Radiance Calibrated Nighttime Lights data. The odds ratio and 95% confidence intervals of depressive symptoms at the high level of outdoor LAN exposure against the low level during the 1 years before the investigation was 1.49 (1.15, 1.92) with p-value for trend < 0.01, and those associated with per interquartile range increase in LAN exposure was 1.22 (1.06, 1.40).

KEYWORDS:

• Outdoor light at night
• depressive symptoms
• veteran
• China

Introduction

Depression is a serious mental illness that can have significant negative impacts on the life of the patient. In addition to causing feelings of sadness, hopelessness, and helplessness, depression can increase the risk of developing other chronic diseases such as heart disease, stroke, hypertension, diabetes, asthma, chronic lung disease, and cancer (Scott et al. 2016; Plana-Ripoll et al. 2019; Momen et al. 2020). Currently, treatment for depression is limited, and at least half of the patients relapse (Burcusa and Iacono 2007). The prevalence of depression has been increasing rapidly worldwide, showing an average growth rate of 13% annually since 1990, with an estimated 280 million people suffering from it in 2019 (Vos et al. 2020). Particularly, cases of major depressive disorders increased by 27.6% in 2020 globally due to the COVID-19 pandemic (Santomauro et al. 2021). In addition to better known risk factors for depression include physical illness, psychosocial stress, and sleep disorders (Alexopoulos 2005; Fiske et al. 2009; Wilkinson et al. 2018), the associations of environmental exposures (e.g. air pollution and noise) and depression are of increasing global concern (Braithwaite et al. 2019; Eze et al. 2020).

Over the past decades, with the rapid development of urbanization and industrialization, exposure to outdoor light at night (LAN) has increased dramatically in many parts of the world (Cauwels et al. 2014). According to the Visible Infrared Imaging Radiometer Suite Day-Night Band data, outdoor LAN is increasing at an annual rate of 2.2% worldwide (Kyba et al. 2017), and over 80% of the global population are suffering from light pollution (Falchi et al. 2016). LAN can disrupt circadian rhythms and sleep (LeGates et al. 2014), which are risk factors for depression (Logan and McClung 2019; Riemann et al. 2020). Evidence from animal studies has shown that night light exposure could led to mood changes and cause depression-like reactions (Bedrosian et al. 2011, 2013; LeGates et al. 2012). However, few epidemiological studies have investigated the relationship between outdoor LAN and depression, with only two studies from high-income countries showing inconsistent results. A cross-sectional study of young and middle-aged Koreans conducted in 2018 showed a significant association between outdoor LAN and depressive symptoms (Min and Min 2018), while another cross-sectional study in 2020 based on Dutch adults (<65 years) reported null results (Helbich et al. 2020). Neither of the two studies considered people aged ≥65 years. Therefore, more studies should focus on the mental health effects of outdoor LAN exposure.

With the elderly population (>60 years) of 264.02 million and a life-time depression prevalence of 7.3% (Lu et al. 2021; Wang 2022), China bears the huge burden of geriatric depression, which is predicted to increase in future (Charlson et al. 2016). Veterans are a special group of population who are more likely to suffer from depression than the general population (Hoerster et al. 2012). Previous studies have shown that separation from family and friends, gruelling military training, bullying, and post-traumatic stress disorder from war during military service could increase the risk of depression in veterans (Xiong et al. 2005; Warner et al. 2007; Inoue et al. 2022). In this study, the relationship between outdoor LAN and depressive symptoms in older adults was examined, using data from the Chinese Veteran Clinical Research (CVCR) platform.

Materials and methods

Study population

The CVCR platform was established by the Chinese People’s Liberation Army General Hospital between 2009 and 2011, which covers 277 veteran communities in 18 cities (see Figure 1). Detailed information about the CVCR platform has been previously reported (Tan et al. 2014). In brief, 10 cities in the developed eastern region of China (including four first-tier cities, five second-tier cities, and one third-tier city) and eight cities in the less developed central and western regions (all second-tier cities) were first selected based on city size and economic status. The sample size was specified as 500 to 1000 for cities with more veteran communities, and 100 to 300 for cities with fewer veteran communities. Eligible veterans in the CVCR platform who met the following inclusion criteria were included in this study: (1) worked in the military before retirement; (2) ≥ 60 years of age; and (3) lived in the veteran community after retirement with no mobility. There are very few female veterans in China. As the sample size for female veterans is too small, only male veterans were included in this study. A total of 6,445 eligible participants from the CVCR platform were finally included in this study.

Figure 1. Locations of veteran communities.

Abbreviations: LAN, light at night.

Depressive symptoms

Depressive symptoms in this study were assessed using the Chinese version of the Center for Epidemiological Studies Depression scale (CES-D). The CES-D was developed by the National Institute of Mental Health and has been widely used to assess levels of depression in community populations (Radloff 1977; Park and Yu 2021). The Chinese version of the CES-D has shown high degrees of reliability and validity in Chinese populations (Rankin et al. 1993; Zhang et al. 2010, 2012). The CES-D consists of 5 dimensions (20 items), including depressed mood, positive mood, somatic symptoms and activity delay, and interpersonal relationships. The weekly frequency of each symptom was scored according to a four-point Likert scale: 0 = less than one day, 1 = 1–2 days, 2 = 3–4 days, 3 = 5–7 days. Face-to-face interview was conducted for each participant by well-trained medical workers. Participants with a total score > 15 indicated depressive symptoms (Wu et al. 1989).

Outdoor LAN assessment

We used data from the US Defense Meteorological Satellite Program’s Operational Linescan System (DMSP-OLS) to estimate participants’ outdoor LAN exposure (https://www.ngdc.noaa.gov/eog/dmsp/download_radcal.html). By removing the effects of sunlight, moonlight, clouds, and lights from transient events from satellite images taken by DMSP-OLS, the National Oceanic and Atmospheric Administration’s Earth Observation Group released the DMSP-OLS Nighttime Lights Time Series with a resolution of 30 arc second (Zhao et al. 2019). However, this data product has small range of LAN value, and bright areas beyond the measurement range are defined as the highest value by default, which results in loss of data details in these bright areas (Zhao et al. 2019). To address the “saturation” problem, the National Geophysical Data Centre (NGDC) calibrated the DMSP-OLS Nighttime Lights Time Series by combining images at low, medium, and high fixed gain levels and released Global Radiance Calibrated Nighttime Lights (Elvidge et al. 1999). In this study, Global Radiance Calibrated Nighttime Lights was used to estimate outdoor LAN exposure for each participant based on their community address (latitude and longitude). We considered participants’ average level of outdoor LAN exposure during the 1 year before investigation in the main analyses, according to previous works (Braithwaite et al. 2019; Paksarian et al. 2020; Zhou et al. 2023). To ensure comparability across years and satellites, we used the corrected nighttime lights data using interannual and interplanetary calibration coefficients provided by the NGDC (Hsu et al. 2015). As the NGDC released nightlight data only for years of 1996, 1999, 2000, 2003, 2004, 2006, and 2010, missing outdoor LAN data were assigned to adjacent years.

Covariates

According to literature, a range of covariates were considered in our analyses (Cole and Dendukuri 2003; Alexopoulos 2005; Fiske et al. 2009; Wilkinson et al. 2018; Braithwaite et al. 2019). Data on participants’ socio-demographic characteristics and behavioral factors were collected by questionnaires, including gender (male or female), age, educational attainment (years of education ≤ 9 or years of education > 9), family history of mental illness (yes or no), and marital status (married or unmarried). Behavioral factors included regular physical activities (yes or no), regular social activities (yes or no), smoking (current, former, or never), and drinking (>2 times/week, ≤ 2 times/week, quit or never). Regular physical activity was defined as exercising ≥30 minutes per day at an intensity ≥ walking, and regular social activity was defined as participating in organized activities involving social contact. In addition, information on participants’ medical history was extracted from hospital records, including insomnia (yes or no), hypertension (yes or no), stroke (yes or no), and diabetes mellitus (yes or no). Additionally, particulate matter with an aerodynamic diameter ≤2.5 μm (PM_2.5) (low or high), which is potentially associated with depression (Braithwaite et al. 2019), was also included in the analyses. A machine learning method (random forest model) was used to estimate participants’ PM_2.5 exposure during the 1 year before the survey using ground-level monitoring data, satellite-based aerosol optical depth, and other spatiotemporal information. More details of PM_2.5 estimation were previously reported (Chen et al. 2021).

Statistical analysis

We used a mixed-effects logistic regression model to examine the association between long-term outdoor LAN exposure and depressive symptoms. A crude model was first developed with only including outdoor LAN exposure and the random-effect term. We then developed an adjusted model by further including a range of potential confounders (Alexopoulos 2005; Fiske et al. 2009; Min and Min 2018; Wilkinson et al. 2018) including age, years of education, regular physical activities, regular social activities, smoking, drinking, marital status, family history of mental illness, history of illness (insomnia, hypertension, stroke, diabetes mellitus), and PM_2.5. Due to the non-linear association between age and depression (Arias et al. 2021), age was included as a smoothing term using natural cubic splines with 4 degrees of freedom in the adjusted model. To control for potential spatial clustering, the region of each participant’s community (East China, Central China, and West China) was included as a random-effect term in all models. Results were expressed as odds ratio (OR) and 95% confidence interval (95%CI) associated with per interquartile range (IQR) increase in outdoor LAN exposure or as ORs and 95%CIs at the middle or high level against the low level of LAN exposure (LAN exposure was divided into three categories according to the tertile). Analyses were also stratified by age, years of education, regular physical activities, regular social activities, smoking, drinking, and marital status. Sensitivity analyses were performed by using other LAN exposure periods, including the 2 and 3 years before the investigation. All statistical analyses were performed with R software (version 4.2.0).

Results

A total of 6,445 male veterans from 18 cities in China were included in this study, of whom 440 (6.8%) suffered from depressive symptoms. The basic demographic characteristics of the participants are shown in Table 1. The mean (SD) age of the participants was 82.1 (4.0) years. Veterans were more likely to suffer from depressive symptoms, who were older, less educated, had no regular physical or social activities, drank alcohol, and had a history of insomnia, hypertension, or stroke. Other factors (smoking, marital status, history of diabetes mellitus, PM_2.5 exposure) were equally distributed between the two groups. Table 2 presents a summary of participants’ exposure to outdoor LAN during the 1 year before the investigation. The median level of outdoor LAN exposure was 65.43 nW/cm²/sr and the IQR was 53.42 nW/cm²/sr.

Table 1. Demographic characteristics of depressive symptoms cases and controls.

Variables	Overall		Depressive symptoms cases		Controls		p-value
	N	%	N	%	N	%
Age (mean ± SD)^a	82.1	4.0	83.0	3.3	82.1	4.0	<.001
Years of education^b							.01
≤9	4195	65.1	312	70.9	3883	64.7
>9	2250	34.9	128	29.1	2122	35.3
Regular physical activities^c							<.001
Yes	5525	85.7	287	65.2	5235	87.2
No	920	14.3	153	34.8	767	12.8
Regular social activities^d							<.001
Yes	2427	37.7	131	39.8	2296	38.2
No	4018	62.3	309	70.2	3709	61.8
Smoking							.03
Current smoker	538	8.3	32	7.3	506	8.4
Ex-smoker	2271	35.2	180	40.9	2091	34.8
Non-smoker	3636	56.4	228	51.8	3408	56.8
Drinking							.001
>2 times/week	497	7.7	24	5.5	473	7.9
≤2 times/week	1592	24.7	100	22.7	1492	24.8
Quit	1051	16.3	101	23.0	950	15.8
Never	3305	51.3	215	48.9	3090	51.5
Marital status							.22
Unmarried	755	11.7	60	13.6	695	11.6
Married	5690	88.3	380	86.4	5310	88.4
family history of mental illness							1.00
Yes	70	1.1	5	1.1	65	1.1
No	6375	98.9	435	98.9	5940	98.9
Insomnia							<.001
Yes	1103	17.1	157	35.7	946	15.8
No	5342	82.9	283	64.3	5059	84.2
Hypertension							0.04
Yes	4374	67.9	319	72.5	4055	67.5
No	2071	32.1	121	27.5	1950	32.5
Stroke							<.001
Yes	1286	20.0	141	32.0	1145	19.1
No	5159	80.0	299	68.0	4860	80.9
Diabetes mellitus							0.02
Yes	1770	27.5	142	32.3	1628	27.1
No	4675	72.5	298	67.7	4377	72.9
PM2.5 (median [IQR])^e	55.0	[47.3, 68.1]	55.5	[47.0, 65.4]	55.0	[47.3, 68.1]	1.00
Region							0.23
East China	4902	76.1	348	79.1	4554	75.8
Central China	562	8.7	37	8.4	525	8.7
West China	981	15.2	55	12.5	926	15.4
Total	6445	100.0	440	6.8	6005	93.2

Abbreviations: SD, standard deviation; PM_2.5, particulate matter with an aerodynamic diameter ≤2.5 μm; IQR, interquartile range.

^aStudent’s t-test was used to examine the difference between veterans with and without depression as the variable was normal.

^bChi-squares test was used to examine the difference between veterans with and without depression.

^cRegular physical activity was defined as exercising at least 30 minutes per day at an intensity equal to or greater than walking.

^dRegular social activity was defined as participating in organized activities involving social contact.

^eKruskal-Wallis test was used to examine the difference between veterans with and without depression as the variable was not normal.

Table 2. A summary of participants’ exposure to outdoor LAN (nW/cm²/sr) during the 1 year before the investigation.

Variables	Overall		Depressive symptoms cases		Controls
	Median	IQR	Median	IQR	Median	IQR
Age
<80	65.77	55.45	79.67	45.46	65.77	56.45
≥80	65.43	52.08	72.17	56.45	65.43	51.75
Years of education
≤9	64.67	51.75	70.89	53.14	64.41	51.59
>9	66.68	56.45	83.90	54.64	66.68	56.61
Regular physical activities
Yes	65.77	53.29	77.23	53.42	65.43	53.29
No	64.40	51.70	70.89	54.68	64.31	51.55
Regular social activities
Yes	66.88	52.50	85.27	52.70	65.77	51.43
No	64.67	53.67	86.64	52.24	66.16	50.24
Smoking
Yes	62.22	58.39	82.93	49.66	61.13	59.27
No	65.77	53.29	74.16	54.68	65.43	52.87
Drinking
Yes	65.43	53.05	77.53	63.26	65.43	52.34
No	65.43	53.29	74.16	50.16	65.43	53.29
Marital status
Unmarried	64.41	52.34	70.27	59.86	64.41	52.34
Married	65.43	53.29	77.23	55.75	65.43	53.42
Total	65.43	53.42	75.22	53.41	65.43	52.59

Abbreviations: IQR, interquartile range.

The results of the crude and adjusted models are shown in Table 3. Brighter outdoor LAN exposure was significantly associated with an increased risk of depressive symptoms. In the adjusted model, the ORs and 95%CIs at the middle and high levels were 1.14 (0.88, 1.47), and 1.49 (1.15, 1.92), respectively, against the low level of average outdoor LAN exposure during the 1 year before the investigation, and the ORs showed an increasing trend (p-value for trend < .01). When outdoor LAN exposure was considered as a continuous variable in the model, OR (95%CI) associated with per IQR (53.42 nW/cm²/sr) increase in outdoor LAN was 1.22 (1.06, 1.40). The results of the stratified analysis are shown in Figure 2. The OR of depressive symptoms associated with per IQR increase in outdoor LAN exposure was significantly higher among those who participated in regular social activities (1.60 [1.25, 2.04]) than among those who did not (1.07 [0.90, 1.28]).

Figure 2. The ORs and 95%CIs of depressive symptoms for stratified analyses associated with per IQR increase in outdoor LAN exposure during the 1 year before the investigation.

Abbreviations: OR, odds ratio; CI, confidence interval.

Models were adjusted for age, years of education, family history of mental illness, physical activities, social activities, smoking, drinking, marital status, insomnia, PM_2.5, and history of diabetes, hypertension, stroke, and region of each community (East China, West China, Central China) as a random-effect term. Age was included in the adjusted model as a smooth term with 4 degrees of freedom. PM_2.5 was divided into low and high groups according to the median.

Table 3. The ORs and 95%CIs of depressive symptoms associated with exposure to outdoor LAN during the 1 year before the investigation in crude and adjusted models.

Outdoor LAN	OR (95%CI)
	Crude model ^a	Adjusted model ^b
Low level (median 30.75 nW/cm^2/sr)	Reference	Reference
Middle level (median 64.67 nW/cm^2/sr)	1.17 (0.91, 1.50)	1.14 (0.88, 1.47)
High level (median 107.73 nW/cm^2/sr)	1.34 (1.05, 1.72)	1.49 (1.15, 1.92)
P for trend^c	0.02	<.01
Continuous LAN (per IQR increase ^d)	1.15 (1.00, 1.32)	1.22 (1.06, 1.40)

Abbreviations: LAN, light at night; OR, odds ratio; CI, confidence interval; IQR, interquartile range.

^aModels included outdoor LAN exposure and region of each community (East China, Central China, and West China) as a random-effect term.

^bModels were additionally adjusted for age, years of education, family history of mental illness, physical activities, social activities, smoking, drinking, marital status, insomnia, PM_2.5, and history of diabetes, hypertension, stroke, and region of each community (East China, West China, Central China) as a random-effect term. Age was included in the adjusted model as a smooth term with 4 degrees of freedom. PM_2.5 was divided into low and high groups according to the median.

^cTest for the trend is based on the median value for each level.

^dIQR = 53.42 nW/cm²/sr.

Figure 3 presents the results of the sensitivity analyses. Significant associations were observed between 2 and 3-years mean LAN exposure before the investigation and depressive symptoms, with the ORs (95%CIs) of 1.24 (1.06, 1.44) and 1.25 (1.07, 1.47) associated with per IQR (53.42 nW/cm²/sr) increase in LAN. It indicated that the association between outdoor LAN exposure and depressive symptoms was robust to different exposure periods.

Figure 3. The ORs and 95%CIs of depressive symptoms at different levels of LAN exposure against the lowest level and those associated with per IQR (53.42 nW/cm²/sr) increase in LAN during different exposure periods.

Abbreviations: OR, odds ratio; CI, confidence interval.

Models were adjusted for age, years of education, family history of mental illness, physical activities, social activities, smoking, drinking, marital status, insomnia, PM_2.5, and history of diabetes, hypertension, stroke, and region of each community (East China, West China, Central China) as a random-effect term. Age was included in the adjusted model as a smooth term with 4 degrees of freedom. PM_2.5 was divided into low and high groups according to the median.

Discussion

This multi-city study revealed that exposure to brighter outdoor LAN was significantly associated with higher risks of depressive symptoms in male veterans of China. To the best of our knowledge, this study is the first study in China to examine the relationship between long-term outdoor LAN exposure and depressive symptoms in older adults.

In recent years, a growing body of studies have linked outdoor LAN exposure to various adverse outcomes, such as cardiovascular disease, cancer, psychiatric disorders, and sleep disorders (Kloog et al. 2009; Paksarian et al. 2020; Sun et al. 2021; Zhang et al. 2021; Xie et al. 2022). However, studies on the relationship between outdoor LAN exposure and depression are limited and the results of existing studies are inconsistent. In a cross-sectional study of 113,119 Korean residents aged 20–59 years, it was reported a positive association between the level of outdoor LAN exposure and the risk of depressive symptoms, with residents with the highest LAN exposure having 1.29 times (95%CI: 1.15–1.46) higher risk of depressive symptoms than those with the lowest LAN exposure (Min and Min 2018). In addition, a cohort study of 863 Japanese older adults showed that LAN in the bedroom was associated with a higher risk of depression with an OR (95% CI) of 1.89 (1.13, 3.14) (Obayashi et al. 2018). However, a cross-sectional study from the Netherlands reported no effect of outdoor LAN exposure on the prevalence of depressive symptoms (Helbich et al. 2020). The difference in results of our study with previous studies are mainly due to different socio-demographic features, especially the disparities between veterans and the general populace. Military service brings severe psychological stress and trauma to veterans, such as leaving their families, being bullied, witnessing war, and suffering physical injuries (Warner et al. 2007; Sayer et al. 2014). Additionally, the transition to civilian life after discharge also comes with challenges, including employment challenges, social network reconstruction, and cultural differences between the civil and military sectors (Sayer et al. 2014; Elnitsky et al. 2017). The aforementioned factors may make veterans more vulnerable to mental illnesses (Seal et al. 2009). Our study provides epidemiological evidence for the relationship between outdoor LAN and depressive symptoms in low- and middle-income countries. As industrial development and technological progress, outdoor LAN is increasing dramatically worldwide (Falchi et al. 2016). Given the adverse health effects of outdoor LAN, burden of outdoor night light related disease is increasing, which makes effective management of outdoor LAN and reducing the exposure urgently in need.

The results of stratified analyses indicated that participants involving regular social activities were more susceptible to the adverse impact of outdoor LAN exposure on depressive symptoms. This may be explained by the fact that a lower baseline risk would highlight the effect of risk factor exposure (Tan et al. 2021), as people who participate in social activities have a lower risk of depression (Min et al. 2016).

Although the biological mechanisms underlying the association between outdoor LAN and depression are unclear, current evidence suggests that LAN can affect mood directly and indirectly. Animal studies have shown that nighttime light could affect mood by suppressing melatonin secretion and altering the expression of clock genes to disrupt circadian rhythms and sleep (Vandewalle et al. 2010; LeGates et al. 2014; Bedrosian and Nelson 2017; Fernandez et al. 2018; Maruani and Geoffroy 2022). Moreover, light can directly affect mood without causing changes in circadian rhythms and sleep (LeGates et al. 2012; Fernandez et al. 2018; An et al. 2020). Nighttime light can induce depressive-like behaviors in mice via a neural pathway that begins with the intrinsically photosensitive retinal ganglion cells (ipRGC), continues through the dorsal perihabenular nucleus (dpHb), and ultimately reaches the nucleus accumbens (NAc) (An et al. 2020).

Our study has several strengths. First, the participants in this study were veterans within the CVCR platform. They had comparable work experiences, socioeconomic conditions, and residential environments, which may help with controlling for potential confounding effects of these factors. The veteran communities included in this study had a well-established health management system, which can provide detailed information on participants’ health status and disease history. An additional advantage is that 277 veteran communities from 18 cities in China were selected for this study using a multi-stage stratified sampling method, and the potential spatial clustering was controlled, which contributes to representativeness and robustness of the results.

Some limitations of our study are worth noting. Firstly, due to the cross-sectional design and unavailability of exact onset time of disease, we are unable to examine the causal relationship between outdoor LAN and depressive symptoms. Second, indoor use of curtains and room lighting may affect individual outdoor LAN exposure. Due to unavailability of such information, we are not able to control factors related to indoor living environment in the analyses. In addition, as the LAN data were not available in selected years during the study period, we used adjacent year’s data to fill in the gaps, which may have impacts on the results. However, the LAN data do not change substantially during a short-period of time (Kyba et al. 2017).

Conclusions

In this study, we found that excess outdoor LAN exposure increased the risk of depressive symptoms in male veterans. Given the increasing medical burden of depression in the elderly, controlling outdoor LAN exposure is essential. To mitigate over-illumination, it is recommended that policymakers impose limits on the amount and brightness of urban lighting (Falchi et al. 2011). At the individual level, effective measures should be taken including reducing outdoor activities at night and using curtains indoors.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the Special Research Project on Health Care, Health Sector of the General Logistics Department of Peoples Liberation Army [project numbers: 07BJZ04, 10BJZ19, 11BJZ09, and 12BJZ46].