Invited Commentary: “Bedroom Light Exposure at Night and the Incidence of Depressive Symptoms: A Longitudinal Study of the HEIJO-KYO Cohort”

Abstract

In modern society, we are increasingly disconnected from natural light/dark cycles and beset by round-the-clock exposure to artificial light. Light has powerful effects on physical and mental health, in part via the circadian system, and thus the timing of light exposure dictates whether it is helpful or harmful. In their compelling paper, Obayashi et al. (Am J Epidemiol. 2018;187(3):427–434.) offer evidence that light at night can prospectively predict an elevated incidence of depressive symptoms in older adults. Strengths of the study include the longitudinal design and direct, objective assessment of light levels, as well as accounting for multiple plausible confounders during analyses. Follow-up studies should address the study’s limitations, including reliance on a global self-report of sleep quality and a 2-night assessment of light exposure that may not reliably represent typical light exposure. In addition, experimental studies including physiological circadian measures will be necessary to determine whether the light effects on depression are mediated through the circadian system or are so-called “direct” effects of light. In any case, these exciting findings could inform novel approaches to preventing depressive disorders in older adults.

Beginning with the Industrial Revolution and continuing into modern society, we have experienced striking changes in our exposure to light (1). More and more, we are disconnected from natural light/dark cycles and beset by round-the clock exposure to artificial light, not only from overhead lighting but from our increasingly pervasive electronic devices. In contrast, the circadian clocks that dictate critically important internal time-keeping functions—clocks that are evolutionarily conserved from simple organisms up through humans (2)—evolved in accordance with the natural light/dark cycle (3). Perhaps unsurprisingly, there is an increasing awareness that light has powerful effects on our physical and mental health (4). Furthermore, it is not just the overall amount of light exposure, but the timing of light exposure that dictates whether light is helpful or harmful. Light has dramatically different effects depending on relative internal (circadian) time in which it is received (5), and it can thus be therapeutic (e.g., bright light therapy for circadian rhythm sleep disorders (6) and an increasing range of psychiatric disorders(7)) or harmful (e.g., carcinogenic light at night in the context of shift work (8)). In their compelling paper in this issue of the Journal, Obayashi et al. (9) offer evidence that more subtle exposure to light at the wrong time (at night) may lead to considerable harm, in this case an elevated incidence of depressive symptoms.

A substantive prior literature has linked shift work to a range of adverse health outcomes, including but not limited to cancer, cardiovascular disease, mood disorders, and substance abuse (10–12), with evidence indicating that the exposure to light at night (LAN) is conferring at least some of the elevated risk (8). A newer and growing literature suggests that non–shift worker populations also may be affected by LAN (13, 14); however, most of this prior work has had significant methodological limitations, including cross-sectional designs and self-report or indirect assessments of light exposure (15–20). Obayashi et al. is a notable addition to the literature, providing first evidence of a prospective link between LAN and depression in older adults. Specifically, among a group of older adults whose bedroom light exceeded 5 lux on average (something between moonlight and an illuminated city street) (21), there was a significantly higher risk of developing depressive symptoms (≥6 on the Geriatric Depression Scale) at follow-up (median 2 years). Secondary analyses demonstrated that an hour or more above the 5-lux threshold was associated with even higher risk of subsequent depression.

The study had some obvious strengths, most notably including the use of a longitudinal design and direct, objective assessment of light levels via a sensor attached to the bed headboard. The prospective analysis reduced the possibility that increased LAN is simply a reflection of altered behaviors related to depression, such as extended time in bed that accordingly increases LAN because the individual retires to bed during daylight hours and remains in bed well after dawn. The headboard-mounted sensor reduced noise inherent to using light sensors on wrist actigraphs (which can be blocked by sleeves or bedding) and bias inherent to self-report. Furthermore, the authors accounted for a number of plausible confounders, including age, sex, body mass index, economic status, hypertension, diabetes, and sleep quality. The study also had some notable limitations that point towards important next steps. The use of a self-report and retrospective measure of sleep quality (the widely-used Pittsburgh Sleep Quality Index) precluded determining whether sleep disruption and extended wakefulness at night (and perhaps open eyes) moderates the effect of LAN. The reliance on 2 nights’ assessment of light exposure raises questions of whether these data reflected typical LAN exposure for these participants. Without physiological measures of circadian phase, it is unknown whether circadian mechanisms were implicated in the observed effect of LAN on depressive symptoms. Finally, as Obayashi et al. note, experimental manipulations will be necessary for determination of a causal link between LAN and depressive symptoms.

Such follow-up experimental studies (and others) are warranted to further investigate the mechanism through which LAN would influence depressive symptoms. Decades of reports have demonstrated that mood disorders are often accompanied by circadian disturbances (22–24), and although the field has yet to converge on a clear mechanism, a number of papers have suggested that misalignment between the internal clock and the behavioral sleep-wake schedule is linked to the extent of depressive symptomatology (25–29). Furthermore, strict laboratory-based chronobiological paradigms (i.e., constant routine and forced desynchrony) have demonstrated that mood (particularly positive affect) is modulated by a combination of circadian timing and the time spent awake (30, 31). The relevance of circadian mechanisms to the present findings would be clearer with the inclusion of physiological circadian measures. Additionally, the light levels were quite low (e.g., 20 lux is comparable to the light levels outside at twilight), with questionable effects on circadian physiology. That said, light history is an important determinant of the sensitivity to light exposure (32), and it is possible that individuals in the study (particularly those more prone to depression) were exposed to lower daytime light levels. In addition, Obayashi et al. cite both human work (33) and animal studies (34, 35) demonstrating that similarly low light levels were sufficient to suppress melatonin and induce mood effects, respectively. Follow-up studies should include circadian measures that can allow assessment of any melatonin suppression and/or phase shifts in endogenous rhythms (i.e., melatonin, core body temperature) that would indicate a circadian effect.

Notably, the mechanism may not be a circadian one, given increasing evidence of so-called “direct” effects of light on mood. Recent animal studies have indicated that light can affect mood-related behaviors and neural circuitry relevant to mood without affecting obvious sleep and circadian rhythm measures (36). Even without observable sleep/circadian changes, however, circadian mechanisms may still be involved, for example, via a sensitivity to light’s mood-altering effects that varies across the 24-hour day or via circadian genotypes and/or phenotypes conferring differential vulnerability to light (37). The corticostriatal circuits central to reward processing and heavily implicated in mood disorders show evidence of daily rhythms, both in glucose metabolism and in reactivity to monetary reward (38–40), and might well be differentially sensitive to the effects of light depending on time of day. In a recent investigation of a mouse model deficient in a specific circadian gene (PERIOD3), LAN increased corticosterone and hippocampal expression of brain-derived neurotrophic factor in both wild-type and Per3^−/− mice, but only the Per3^−/− mice also showed signs of anhedonia (reduced preference for sucrose) (41). Meanwhile, in adolescent humans, the degree of eveningness preference (consistent with a later circadian phase) not only correlates with greater depression but has also been linked to the extent of exposure to LAN (42), hinting that the LAN-depression association may not be limited to older adults. Beyond developmental considerations, other plausible but currently indeterminate factors include individual differences in the retinal sensitivity to light (such as in seasonal depression (43)) and the chronicity of exposure. (Does the 2 nights of light exposure in the present study reflect a years-long pattern or a relatively ephemeral pattern immediately preceding subsequent depressive episodes?)

In conclusion, these intriguing prospective data raise a number of questions about the link between LAN and depression. Given that the likelihood of LAN exposure appears to be increasing for all of us, and the compelling findings of Obayashi et al., further efforts towards understanding this phenomenon could portend great public health benefit. There appears to be little to be lost and possibly much to be gained by taking steps towards prevention, such as encouraging older individuals to keep their bedrooms as dark as possible at night.

Abbreviations

LAN

light at night