Are we still in the dark? A systematic review on personal daily light exposure, sleep-wake rhythm, and mood in healthy adults from the general population

Insufficient light exposure is assumed to be related to a wide array of health problems, though few studies focus on the role of whole-day light exposure in the habitual setting in the development of these health problems. The current review aims to describe the association between personal light exposure in the habitual setting and sleep-wake rhythm and mood in healthy adults from the general population. Five databases (Embase, Medline Epub, Web of Science, PsycINFO, and Google Scholar) were searched in June 2019. The inclusion criteria included: assessment directly of light exposure on the participants for at least one full day; reporting on both individual personal light exposure and outcomes. The quality of the papers was assessed using the Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies of the National Heart, Lung and Blood Institute. The current review followed the PRISMA guidelines. In total, 8140 papers were identified in the database search. Twenty-five papers were eventually included in this review. All included studies were cross-sectional, and individual light exposure was usually measured with a wrist-worn device. Five studies received a “good” quality rating, 16 received a “fair” rating, and the remaining 4 a “poor” quality rating. The overall quality of the included studies was considered low because of the lack of intervention studies and the fact that light exposure was measured on the wrist. Given the low quality of the included studies, the current review can only provide a first exploration on the association between light exposure and sleep-wake rhythm and mood in healthy adults from the general population. Limited evidence is presented for a positive relationship between the amount and timing of light exposure on the one hand and rest-activity rhythm and some estimates of sleep architecture on the other. The evidence on an association between light exposure and circadian phase, sleep estimates, sleep quality, and mood is conflicting. Data from intervention studies are needed to gain insight into the causal mechanism of the relationship between light exposure and sleep-wake rhythm and mood. © 2021 The Authors. Published by Elsevier Inc. on behalf of National Sleep Foundation. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/)


Introduction
Light is as important for perceiving the world as it is for regulating physiological and behavioral rhythms. These follow a circadian rhythm, a rhythm of approximately 24 hours. Circadian rhythms are regulated by the biological clock, which is located in the suprachiasmatic nucleus in the hypothalamus. Circadian rhythms synchronize to the external 24h rhythm using external cues called "Zeitgebers," of which light is the strongest. 1 Alignment of the circadian rhythm with external day-night rhythm is considered essential for the regulation of the sleep-wake rhythm and mood. [2][3][4] Insufficient, or badly timed, exposure to light can ultimately result in desynchronization of the sleep-wake rhythm and sleep problems, which in turn can result in mood problems. 3 Misalignment of circadian rhythms 2,5 and disturbances of the sleepwake rhythm 6 and mood 7 are each associated with poor health outcomes.
Research in populations exposed to light patterns that are extremely deviated from the natural dark-light cycle (e.g. shiftwork, jetlag, and a late chronotype) has provided knowledge on the relationship between light exposure, sleep-wake rhythm mood, and other health complaints. For instance, traveling through time zones has been associated with disruption of the circadian rhythm, 8 whereas shiftwork is associated with sleep problems and lower sleep quality, 9 depressed mood, 10,11 cardiovascular diseases, gastro-intestinal and metabolic disorders, 12 as well as cancer. [13][14][15] The exposure to extremely deviating light patterns is assumed to play a role in the development of these health problems. 11 To date, evidence on the relationship between light, sleep-wake rhythm and mood is provided by both fundamental studies in humans and studies in populations with extreme deviating light exposure patterns. For the general population in everyday living conditions this relationship is less clear. Associations between light exposure and health in populations prone to extreme deviating light patterns are not necessarily generalizable to the general population. Although not as extremely deviating, the general population might be exposed to suboptimal light as well; for instance due to little bright light exposure, [16][17][18] poorly lit homes and workspaces, [18][19][20] or evening light exposure using multimedia devices. 21 So far, we are lacking an overview of the evidence on the relationship between daily light exposure in the habitual setting and sleep-wake rhythm, and mood in the general population.
The aim of the current systematic literature review is to provide an overview of the literature on the association of personal light exposure in the everyday (habitual) setting, with sleep-wake rhythm, and mood in healthy adults from the general population. The current review will focus on personal light exposure as measured directly on the participants, as this is considered more reliable than a proxy such as light exposure measurements in the environment. 22 No restrictions will be made with regard to the features of light exposure; the timing, amount, and spectral distribution will all be described. Also, the outcomes in terms of sleep-wake rhythm and mood will be explored broadly and without restrictions. This way, we aim to determine the potential impact of personal daily light exposure on sleepwake rhythm and mood.

Methods
The PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines 23 were followed during this review. This review was registered in the Prospero register under number CRD42016039107.

Search strategy
The search was performed within multiple electronic databases in order to reduce the risk of missing eligible articles. Embase, Medline, Psychinfo, Web of Science, and Google Scholar were searched on June 14, 2019. The database-specific search codes used are shown in Appendix A. In addition, the reference lists in the included papers were examined for further relevant papers.

Definitions and inclusion criteria
The aim of the review is to study the association of personal light exposure in the habitual setting during the day with sleep-wake rhythm and mood in the general population. Personal light exposure refers to the light exposure measured on the participant directly as this is considered more reliable than indirect measurements of light exposure such as light measurements in the environment. 22 Light exposure could be both artificial light and sunlight. All features of light exposure were included; the timing, amount, and spectral distribution will all be described. The habitual setting is defined to include all real-life settings. Studies conducted in a laboratory or experimental setting were excluded. Real-life intervention studies that included baseline analyses of the association between light exposure and outcomes were included. With regard to the population, we aimed for healthy adults from the general population who were not likely to be exposed to extreme deviating light exposure patterns, therefore excluding shift workers, studies of chronotypes, and jetlag. No criteria were formulated for the outcome variables of sleep-wake rhythm, thus 24 hour activity-rest rhythm, sleep estimates or sleep problems could all be included. "Mood" was used as an umbrella term to capture all possible facets of the term; both states and traits (affect vs. temperament) could be included, as well as mood complaints.
Studies had to meet the following inclusion criteria: published in the English language, study sample of healthy adults aged 18 from the general population; light exposure was measured directly on the participant for at least one full waking day; the study took place in a habitual setting; sleep-wake rhythm and/or mood, and analyses of the relationship between habitual daily light exposure and sleepwake rhythm and/or mood were reported.
Studies were excluded if participants were likely to have deviating light patterns (shiftwork, jetlag, chronotype), participants had preexisting sleep-or mood disorders or other known physical or mental conditions, if the full text was not available through the medical libraries or if the text was a conference abstract. Studies conducted in the controlled environment of a laboratory were also excluded.

Selection process
The selection of papers was carried out by the first and second authors. After the electronic database search, the titles and abstracts were screened for eligibility. Then the full text was read of the eligible papers. Disagreement between the 2 authors was resolved through consensus discussions. The bibliography of selected papers was screened for possible relevant papers that were not included in the electronic database search. These papers were screened in the same manner as papers from the electronic database search (Fig. 1).

Data extraction and management
In the data collection phase, all results were extracted of analyses of the relationship between light exposure and the outcome measures sleep-wake rhythm and mood.

Quality assessments
Selected studies were assessed for methodological quality using the Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies of the National Heart, Lung and Blood Institute. 24 The Quality Assessment Tool consists of 14 items that can be scored as "yes," "no," "not applicable" or "not recorded". Each "yes" was rewarded with one point; studies could score a maximum of 14 points.
Based on the quality assessment, an overall rating of "poor," "fair" or "good" was determined. Studies that scored 9 points or more received an overall "good" rating; however, if only basic analyses (such as correlations) were performed the study was assigned a "fair" overall rating. Studies with 8 points with advanced analysis (such as regression analyses) received a "good" rating. Studies with 6-8 points received a "fair" rating, unless they had a small sample size (N < 30); those studies were given a "poor" rating. Studies with 5 points or fewer were all rated as "poor." The quality assessment was done by the first author, who assessed all the included studies and determined the overall rating for the quality of the study. Next, the findings were presented to the second author to check whether the quality assessment of all the studies had been performed in a consistent manner. Disagreements were resolved through a consensus discussion.
The current review itself was not assessed on quality, as to date no appropriate tool exists to do a quality assessment of reviews that include merely observational studies.

Selection of studies
The search for studies on the association between daily light exposure and sleep-wake rhythm and mood generated 8140 unique articles (Fig. 1). Based on title-abstract screening, 303 were considered eligible. After reading the full text of these articles, 23 articles were included. Most of the excluded studies (n = 279) were excluded because they did not report on personal light measurements. After including 3 articles from the reference lists of the included papers, a total of 26 articles were obtained for data extraction. With regard to overlapping populations, 3 papers published results on the same group of post-menopausal women from the Women's Health Initiative. [25][26][27] In 2 cases, there were 2 different papers that reported on the same group of healthy volunteers. [28][29][30][31] For these studies, only the first unique analysis will be reported. Two studies 25,32 described identical populations and data analysis relevant for the current review, therefore one of these-the study of Youngstedt et al (2004) 32 -was excluded.
Of the final 25 articles included, ten focused solely on sleep-wake rhythm, 30,33-41 4 on sleep-wake rhythm and mood, 19,25,27,42 and the remaining 11 studies focused solely on mood. 26,28,29,31,[43][44][45][46][47][48][49] Quality assessment of the included studies Information needed to rate the items of the Quality Assessment Tool was not always reported in the articles, which restricted the proper assessment of the risk of bias. Five studies received an overall "good" rating. 19,34,36,43,49 Sixteen studies were rated as "fair." 25,[27][28][29][30][31]33,35,37,39,41,42,44,[46][47][48] The remaining 4 studies were rated as "poor." 26,38,40,45 No sample size justification was provided for 22 studies, but due to the observational nature of the studies, this was not considered to impact the quality of the studies too much. Twenty studies did not report on the participation rate of eligible persons, which might have resulted in an unrepresentative sample of the target group. Fifteen studies measured the light exposure once or averaged the light exposure over the measurement period, making it not possible to show an effect of changes in light exposure over time, thus resulting in a weaker study design. Fourteen studies measured the light exposure (mostly) at the time of the outcomes rather than prior to the outcomes, which makes it not possible to study a causal relationship between light exposure and the outcomes. Table 1 gives the complete overview of the results of the quality assessment.

Light exposure measurement
Within the 25 studies, 11 different devices were used to measure personal light exposure. Measuring light exposure at eye level is preferred as it is the most reliable way to estimate the amount of light that enters the eye. Three studies measured light exposure at eye level using a device that was attached to glasses that the participants wore. 19,45,46 In one study, the participants wore a lanyard with a light sensor as a necklace 38 ; in another study the light cell was pinned as a brooch. 49 In 19 studies, participants wore an accelerometer with an integrated light cell on their wrist. [25][26][27][28][29][30][31][33][34][35][36][37][39][40][41][42][43][44]47,48 New-generation wrist-worn accelerometers, used primarily to measure sleepwake activity, often come with an integrated light sensor. Although practical, the amount of light exposure measured at wrist-level is less reliable. When compared to eye level, the amount of light exposure measured at wrist level deviates by up to 27%. 50 All studies measured light at least for one complete waking day of the participants, as this was an inclusion criterion. The duration of the measurements varied between 16 hours and 10 days.

Associations between light exposure and outcomes
Based on the reported light exposure outcome variables, a description will be given of the association between the amount (defined as the average intensity of light exposure in lux, unless specified otherwise), duration, and timing of light exposure on the one hand and the outcome variables sleep-wake rhythm, and mood on the other hand. The results will be described starting with the highest quality study included. In the case of significant results, all the test results and p-values for the light exposure variables reported in the papers are presented in both the text and the tables.
We acknowledge the broadness of our research question, but we are interested in determining the potential impact of personal daily light on sleep-wake rhythm, and mood even if we could not be definitive about the specific conditions and effects. Therefore, in order to qualify the results, the conclusions per domain are drawn based on an adaptation of the Cochrane classification of the level of evidence 51 : Strong evidence-consistent findings among multiple high-quality studies; Moderate evidence-consistent findings among multiple fair/lowquality studies and/or one high-quality study; Limited evidence-consistent findings among multiple low-quality studies; Very limited evidence-single low-quality study; Conflicting-inconsistent findings among multiple studies; No evidence-no studies available.

Assessment of sleep-wake rhythm
Outcomes for the sleep-wake rhythm are described within the domains sleep-wake rhythm and sleep quality. The 12 studies on light exposure and sleep-wake rhythm and 7 studies on light exposure and sleep quality are described in Tables 2 and 3. Of these studies, 5 studies solely used actigraphy to measure sleep-wake rhythm. 27,37,39,41,42 One study used both actigraphy and sleep diaries or questionnaires. 25 Four studies focused on measuring melatonin levels or dim light melatonin onset (DLMO) as indicator of the circadian phase of the sleep-wake rhythm. 30,[33][34][35] Wams et al (2017) 36 used polysomnography (PSG) in combination with actigraphy and DLMO measurements.
Sleep quality was measured in 7 studies, of which 4 36,39,40,42 used the well-validated Pittsburgh Sleep Quality index (PSQI). 52 Two actigraphy studies reported the sleep efficiency as a proxy for sleep quality. 27,39 The self-reported PROMIS Sleep disturbance short form for measuring sleep problems in addition to measuring sleep quality was used twice. 40,42 The last 2 studies 19,25 added questions on subjective sleep quality to the questionnaires.

Association between light exposure and sleep-wake rhythms and sleep quality
Because of the broad variety of outcome measures for sleep-wake rhythms and sleep quality, results are grouped per outcome measure.

Grandner et al (2006)
Were key potential confounding variables measured and adjusted statistically for their impact on the relationship between exposure(s) and outcome ( Note. Question 5: Sample size of n < 30 was considered small. Question 6: studies that measured exposure prior to the outcome received a "yes"; if exposure was measured at the same time as or after the outcome, this question was scored "no". Question 7: studies received a "yes" if light exposure was measured for at least 3 days. Studies that scored 9 points or more received an overall "good" rating; however if only basic analyses (such as correlations) were performed the study was assigned a "fair" overall rating. Studies with 8 points that included advanced analysis (such as regression analyses) received a "good" rating. Studies with 6-8 points received a "fair" rating, unless they had a small sample size (N < 30)those studies were assigned a "poor" rating. Studies with 5 or less points were all rated as "poor".  Table 2 Characteristics of included studies that examine the association between personal light exposure and sleep-wake rhythm in the general population      With regard to the timing of light exposure, the fair study found an association between exposure and the amount of light exposure in the morning and phasor magnitudes (F 1,45 = 41,94, p < .0001). 42 In the poor study, a greater amount of light exposure in the morning (r = 0.437, p < .001) and evening (r = 0.304, p < .01) was associated with a more stable rest-activity rhythm. 38 Overall, based on 2 lower-quality studies 38,42 with consistent results, it is concluded that limited evidence is available for a positive relationship between the amount and timing of light exposure and rest-activity rhythms Table 2.
Circadian phase of sleep-wake rhythm One good-quality study found a positive relationship between the amount of light exposure and DLMO (R 2 = 0.23, x 2 (2) = 10.01, p < .01), whereas maximal light exposure in lux was not related to DLMO. 36 Another good-quality study found no relationship between the amount of white light and DLMO, but more exposure to blue light was related to an earlier DLMO (r = À0.46, p = .01). 34 A fair-quality study showed light exposure mesor was not related to timing and mesor of melatonin. 30 Another fair study found that total duration of light exposure above any level of light exposure did not correlate with DLMO. 35 With regard to timing of light exposure, the good-quality study did not find a relationship between timing of light exposure and DLMO. 36 A fair study found a medium to strong (r = 0.49-0.77) positive relationship between the timing of first and last light exposure and DLMO; later exposure to light was related to later DLMO. 35 Another fair study compared DLMO timing in regular and irregular sleepers, and concluded that the 1.7-hour delay of DLMO in irregular sleepers (p < .01) was the result of their delayed timing of light exposure. 33 The last fair study showed later light exposure acrophase was related to lower melatonin mesor (r = À0.43, p = .03) but not to melatonin acrophase. 30 Given the 2 high-quality studies 34,36 that found a positive relationship between the amount of light exposure and DLMO and 2 fairquality studies 30,35 that did not find a significant relationship, the available studies provide conflicting evidence on the association between light exposure and DLMO. Given the one high-quality study 36 that did not find a relationship and 3 fair studies that found a positive relationship between timing of light exposure and DLMO, 30,33,35 the evidence for an effect of the timing of light exposure on DLMO is conflicting too.
Sleep architecture The highest-quality study showed exposure to light of higher intensity was associated with a lower percentage of stage 1 sleep (p = .03), shorter REM sleep duration (R 2 = 0.43, x 2 (2) = 13.90, p < .001), and longer slow wave duration on PSG (R 2 = 0.25, x 2 (2) = 8.86, p < .05). 36 The amount of light exposure was not associated with the percentage of stage 2, 3 and REM sleep, N3 latency, and REM sleep latency. 36 A fair study showed that L5 onset, the start of the 5 hours with the least amount of activity during the night, was earlier when participants were exposed to a higher amount of light during the day (r = À0.23, t(60) = À2.58, p = .012). 41 Sleep fragmentation was not associated with the amount of light exposure in one fair study and one poor study. 38,39 The highest-quality study showed that later timing of first exposure to >10 lux (R 2 = 0.21, x 2 (1) = 5.77, p < .05) and later timing of maximal light exposure (R 2 = 0.36, x 2 (2) = 11.17, p < .01) were associated with a subsequent shorter latency to first REM episode. 36 Timing of light exposure was not associated with the percentage of stage 2, 3 and REM sleep, N3 latency, and REM sleep latency. 36 A fair study found nocturnal activity was not associated with the amount of evening light exposure. 41 A poor study found a higher amount of light exposure in the morning (r = À0.651, p < .001) as well as evening light exposure (r = À0.287, p < .01) was associated with an earlier sleep midpoint. 38 The high-quality study 36 as well as the 2 lower-quality studies 38,41 found evidence for a positive relationship between the amount and  Minutes per day exposed to >1000 lux GDS-score Time exposed to light >1000 lux did not explain the relationship between farming habits and depressive symptoms. Further adjustment for log-transformed time exposed to bright light (1000 lx) and daytime physical activity (model 3 in    timing of light exposure and some estimates of sleep architecture, but due to the broad variety of outcomes this evidence is classified as limited. Sleep timing One high-quality study. 36 and 2 fair-quality studies 41,42 measured sleep onset time (hh:mm, clock time of sleep start); only one fair-quality study found that the amount of light exposure during the day (r = À0.41, t(60) = À3.49, p < .001) and low evening light exposure (r = 0.61, t(60) 5.93, p < .001) were associated with an earlier sleep onset. 41 The high-quality study found sleep offset time (hh:mm, clock time of sleep end) and time in bed (minutes between getting into bed in the evening and out of bed in the morning) not to be associated with the timing or exposure to high light intensity. 36 The high-quality study 36 and one study of fair quality 42 found no relationship between light exposure and sleep timing, whereas one fair-quality study 41 did find a significant relationship. Thus the available studies provide conflicting evidence for a relationship between the amount and timing of light exposure and bedtimes.
Nocturnal sleep duration and daytime napping The only high-quality study found lower light intensities to be related to longer nocturnal sleep duration (R 2 = 0.05, x 2 (1) = 4.83, p < .05) (Wams et al, 2017). The 3 fair-quality studies found the amount of light exposure and nocturnal sleep duration were positively (r = 0.483, p = .03), 39 negatively   25 : r p = À0.20, p < .001) or not 27 associated. Subjective nocturnal sleep duration 25 and duration of daytime napping were not associated with exposure to high light intensities. 25,27 The single high-quality 36 and 3 fair-quality studies 25,27,39 provide conflicting evidence on the relationship between the amount of light and sleep duration.
Sleep onset latency The single high-quality study found light intensity not to be related to sleep onset latency. 36 One fair study found that exposure to higher light intensities was associated with a shorter sleep onset latency (r = À0.29, p < .001), 27 whereas another fair study did not find an association. 39 The high-quality study 36 as well as 2 studies of fair quality 27,39 found no association between timing of light exposure and sleep onset latency. One fair study found high light exposure in the morning to be associated with shorter sleep onset latency in the following night (F 1,15 = 10.43, p = .005). 42 Based on one high-quality study 36 and 3 lower-quality studies, 27,39,42 the evidence on the relationship between amount and timing of light and sleep onset latency is conflicting.
Waking after sleep onset The high-quality study of Wams et al (2017) 36 found that lower maximal light exposure (R 2 = 0.26, p < .05) and earlier timing (R 2 = 0.36, p < .05) of light exposure resulted in fewer awakenings measured using actigraphy, but not when using PSG. Fair-quality studies found a longer duration of light exposure (R 2 = 0.519) 37 and earlier light exposure (Wallace-Guy et al (2002) 27 : r = À0.29, p < . 05) were associated with fewer nocturnal awakenings. In contrast, the fair-quality study of Boubekri et al (2014) 39 did not find an association between the amount and timing of light exposure and wake after sleep onset (minutes). The single high-quality study 36 and 3 fair-quality studies 27,37,39 provide conflicting evidence on the relationship between the amount and timing of light and waking after sleep onset.
Sleep quality ( Table 3) Exposure to higher light intensities was associated with better sleep quality in one high-quality study (F 1;81.1 = 6.84, p = .01) 19 ; the other high-quality study did not find this relationship. 36 Three fairquality studies found exposure to higher light intensities to be associated with better sleep quality   25 39 : p = .05). One poor study found no association between the amount of light and sleep quality. 36,40  One high-quality study found a longer duration of light exposure > 1000 lux (F 1;76.2 = 4.22, p = .04) and >2500 lux (F 1;81.3 = 6.82, p = .01) during the waking day to be positively related to sleep quality. 19 The other high-quality study and a fair-quality study found earlier timing of light exposure to be related to better sleep quality (Wams et al (2017) 36 : t(1)= -2.5, p = .0267; Figueiro et al (2017) 42 : F 1,165 =4.76, p = .031), whereas the other fair study did not find timing to be related to sleep quality. 27 Two fair studies found exposure to higher light intensities to be associated with fewer sleep disturbances   25 : falling asleep; r p = À0.17, p < .005; night waking: r p = À0.18, p < .001; trouble getting back to sleep r p = À0.11, p < .025; Figueiro et al (2017) 42 : F 1,21 = 6.12, p = .022). One study of poor quality found no association between the amount of light and sleep problems. 40 The available studies provide conflicting evidence for the amount and timing of light exposure and sleep quality and the association between higher light intensities and sleep problems. No evidence is presented for the timing of light exposure and sleep problems.
Affect, overall emotional well-being, and quality of life. One study of good quality found the amount of light exposure was associated with positive affect (F 2, 89 = 11.66, p < .0001), 43 but in a poor-quality study these were not found to be associated. 42 The amount of light was associated with more vitality (b = 0.08; p < .01). 46 Other measurements of affect, for instance negative affect, were not related to measures of light exposure. 19,28,40,42,43,46,49 Exposure to a higher amount of light was related to better emotional well-being (r = 0.128, p = .05) and a higher quality of life (r = 0.185, p = .0005), whereas exposure to more light in the morning was associated with quality of life (when corrected for average light exposure, partial F(1400) = 5.760, p = .05, R 2 change=0.013) but not with emotional well-being. 26 Light acrophase was not related to either quality of life or emotional well-being.
Based on the low quality of the studies, the diversity of the outcomes, and the lack of significant results, it is concluded that the available studies provide very limited evidence for an association between the amount of light exposure and affect, emotional wellbeing, and quality of life.

Discussion
This systematic review describes the association between habitual personal light exposure and sleep-wake rhythm, and mood in the general healthy adult population. The 25 articles included in this review mainly focused on the average light intensity in lux of light exposure during the day and the duration and timing of the light exposure. The quality assessment of the 25 included papers revealed a risk of bias in all studies, largely due to gaps in the information reported and because most of the studies were cross-sectional. Limited evidence is presented for a positive relationship between the amount and timing of light exposure on rest-activity rhythm and some estimates of sleep architecture. For the association between light exposure and circadian phase of the sleep-wake rhythm, sleep estimates, sleep quality, and mood, the evidence is conflicting.
The 2 high-quality studies on light exposure and circadian phase of sleep-wake rhythm provided conflicting results, which is not in line with the laboratory studies conducted previously. These show that light sources of just 8 lux can shift the phase of the sleep-wake rhythm. 54,55 Although circadian phase of the sleep-wake rhythm seems to be affected by light exposure within the laboratory setting, the implications for the real world are yet to be determined. Second, there is no consensus on what "an optimal aligned circadian rhythm is" or what a cut-off point is for desynchronization of the rhythms. 55 As all the studies included were cross-sectional and measured circadian phase as a continuum, this review unfortunately cannot give insight into this matter.
Five out of 6 studies (2 of high quality, 3 of fair quality) on sleep quality found a positive relationship between light exposure and sleep quality, while results for several sleep indicators are conflicting. It is hypothesized that the measurements of sleep quality capture an overall association between light exposure and sleep, whereas the different sleep indicators are possibly too specific. Decades of sleep research show a broad variety in sleep preference; some people feel energetic after 8 hours whereas others report a need for 10 hours of sleep. This subjective experience of sleep quality, even more than the objective sleep, is an important criterion in diagnosing sleep problems. 56 Therefore, it is suggested that the results of the review on light exposure and sleep quality might provide more valuable information than the results of the sleep indicators.
The limited evidence for an association between personal light exposure and rest-activity rhythm and sleep architecture does not seem to translate directly to sleep estimates and mood outcomes. The results of the current review were inconclusive for the effect of light exposure on sleep estimates and mood. In these cases, the personal preference, or "chronotype," could provide more insight. Roughly 15% of the population identify as an early chronotype or "lark". Larks are characterized by waking up early in the morning, and falling asleep early in the evening. Another 15% of the population identify as a late chronotype or "owl" and wake up and go to bed later in the day. The remaining 70% have an "intermediate" chronotype. 57,58 Research showed that owls are more prone to depression and anxiety 59,60 and sleep longer 61 than larks. If chronotype is a mediator or confounder for the relationship between light exposure, sleep, and mood, more and earlier light exposure could have a positive relationship with outcomes for larks, and a negative one for owls. It is hypothesized that since chronotype was not taken into account in the included studies, the possible different associations between light exposure and sleep for larks and owls offset one another in the results, resulting in the non-significant results found.
Another explanation for the lack of conclusive results on light exposure and sleep might be provided by the age of the study sample subjects. The aging brain is less sensitive to light exposure, 62 possibly resulting in more light needed to affect sleep. For sleep we included 3 studies with an older study sample, 25,27,37 one with a young sample 36 and 2 with a mixed-age study sample. 39,41 Further inspection of the results of these studies did not provide more insight, as studies with both younger and older subjects showed an association between light exposure and sleep for some measures but not for others.
Finally, the ambiguous results might be explained by the fact that the included studies did not correct for other factors that are known to affect the sleep and mood. Whereas most of the studies corrected for age, they did not take into account other "Zeitgebers" and factors that can affect sleep and mood, like physical activity, working times, diet, or medication use. 63,64 A disadvantage of the included studies is that only a third of the studies measured the light exposure prior to the outcome measurements. Light is shown to have a direct effect on mood and sleep 3,54 in the lab setting, so it is desirable to analyze mood or sleep in respect of the personal light exposure on that same day. Of the 25 papers included in this review, only 5 studies analyzed the data in this manner. Secondly, in order to give a direction to the studied relationship between light exposure, sleep-wake rhythm and mood, it is required that studies measure exposure prior to the outcome and at least at 2 time points. Unfortunately, as all studies were cross-sectional and most just performed correlational analyses, no conclusion can be drawn on the causal relationship between light exposure, sleepwake rhythm and mood.
In line with the above, most of the included studies were not designed to answer the research question of this review. It is hypothesized that some of the included papers reported baseline data of intervention studies. Other papers were by-catch from other studies that happened to have measured both light exposure and outcomes of interest. This was first noticed in the data extraction; studies would report light exposure and mood or sleep outcomes, but no analyses relating these variables. Second, some study populations were part of a bigger cohort in which the wrist-worn accelerometer was used and the light exposure data was analyzed in an exploratory fashion.
Lastly, 19 studies measured personal light exposure using a wristworn light cell, which has been shown to be unreliable. Aarts et al (2017) 50 showed that even within the same wrist-worn devices, the measured light exposure can differ by up to 27% from the actual light exposure. Therefore, measurements of light exposure in the included studies might have been unreliable and might have resulted in the ambiguous results.

Strengths and limitations
Due to the large variation in outcome measures, conducting a meta-analysis was not possible for the current review. In addition, another limitation is that "grey literature" and papers in a language other than English were not included.
One strength of our systematic review is the duplicate study selection and quality assessment, which was performed by 2 authors independently. Second, the current review applied strict inclusion criteria for light exposure measurements and analysis. This way, even though the quality of the included studies was low, this review provides us with more insight into the current evidence on the relationship between personal light exposure, sleep, and mood in the general population.

Future research
Further research should first and foremost be focused on better measurements of personal light exposure. Instead of measuring light exposure on the wrist, it is advised to measure light exposure at eye level or at least chest height as this is more reliable. In addition, measurement of the spectral properties of the light exposure is advised to gain insight into the properties of light that are the most efficient in entraining the circadian rhythm.
In order to gain insight into the causal relationship between light exposure and health, a high-quality, longitudinal intervention study of light exposure, sleep-wake rhythm and mood is needed. In this study, special attention should be given to measuring possible confounders of this relationship, like mental and physical condition, medication use, physical activity, and diet.
Most of the previous work on the association between circadian rhythms and health outcomes was based on populations that are prone to misalignment of the circadian rhythms. The current review provides insight into the relationship between light exposure, sleepwake rhythm and sleep problems in the general population. This review gives grounds for integrating personal light measurements in research on light exposure and health in the populations that are at risk of extreme misalignment of the sleep-wake rhythm, in order to be able to define the mechanism of this relationship more clearly.

Conclusion
The current review aimed to describe the association between personal light exposure in the habitual setting, sleep-wake rhythm, and mood in the general population. Because the quality of the included studies was generally low, this review cannot do more than provide a first exploration of the available literature on this matter. Based on the available studies, we conclude that there is limited evidence for a positive relationship between the amount and timing of light exposure on the one hand and rest-activity rhythms and some estimates of sleep architecture on the other hand. The evidence on the association between light exposure and circadian phase of the sleep-wake rhythm, sleep estimates, sleep quality, and mood is conflicting. High-quality intervention studies are needed to gain insight into the causal mechanism of this relationship.