Although perhaps surprising to those outside the field of architectural lighting, incandescent residential lighting levels are typically lower than those experienced in offices and much lower than those experienced outdoors during the daytime. General ambient light levels on the floor or other horizontal surfaces in a home rarely exceed 300 lux 26. Even for special visual tasks in residences, such as grooming and cooking, vertical light levels likely to be incident on the face or eyes are usually between 50 and 200 lux, and light levels on the visual task rarely exceed 500 lux. Light levels from fluorescent lamps on task areas in commercial and industrial spaces can be two to three times higher, and those outdoors from daylight and sunlight can be 20 to 100 times greater. In buildings illuminated by ceiling luminaries containing fluorescent lamps, which direct light downward, light levels at the cornea are about one-fifth of those measured on a horizontal surface, so in the home, ambient illuminances from incandescent lamps rarely exceed 200 lux at the cornea, and more typically at night are between 10 and 100 lux, based on measurements of vertical illuminances in several residences (Table 127) and based on data obtained with a new instrument designed to record and store (for up to a week at a time) both conventional photopic illuminance and circadian light exposure measurements made at the plane of the eye 28.

Figures 1 and 2 show representative profiles of circadian light exposure for two diurnal women working in a daylighted office during the day, and for two night-shift nurses, respectively, along with expected outdoor light exposures from daylight 26 at the same geographic location as the women represented in the two figures.

As seen in Figure 1, the two diurnal women working indoors were exposed to dim, extended and aperiodic circadian light exposure quite unlike a hypothetical woman working outdoors under a robust light-dark cycle, assumed to be ideal for regulating the circadian system. Note too that the times of light exposure for both of these day-shift women have been extended several hours beyond sunset and that the magnitude and the variation in these nocturnal light exposures are similar to those they experienced during the day. Interestingly as well, light exposure in public spaces from high intensity discharge or fluorescent lamps, such as in gymnasia, for example, are much higher than residential light exposures, and can be experienced for several hours, so these events, if experienced on a regular basis, could possibly be of greater impact on circadian regulation and melatonin production than the ambient lighting from incandescent lamps typically found in residences.

Figure 2 shows similar data to those in Figure 1 but for two female nurses who work during the night shift. These women also experience low light exposures relative to outdoor daytime light levels, but this would be expected because they are asleep during most of the daytime. These night-shift women also seem to have relatively lower light exposures during their waking hours than the day-shift women experienced during their waking hours.

It should be emphasized that the lighting profiles in Figures 1 and 2 are only "snapshots" of light exposure experienced by day- and night-shift workers. Much longer recording periods would likely be necessary to accurately predict circadian response to light because the sensitivity of the circadian system to LAN on a particular night will be affected by a person's light exposure history. Sensitivity to LAN in terms of melatonin suppression has been shown to be higher (i.e., less light is needed to suppress melatonin) after a prior history of exposure to low light levels and lower after a prior history of high light levels 293031. Thus, it probably will be necessary to track several days of light and dark exposures to begin to understand the impact of LAN on nocturnal melatonin suppression for people in their natural environments.

To link architectural LAN to breast cancer risk, the lighting profiles experienced by people, such as those in Figures 1 and 2, ultimately need to be translated into parametric lighting studies with laboratory animals used as models for cancer research. Bullough and colleagues 32, after a literature review and calculations, showed that threshold levels of light (quantity, spectrum, distribution and duration) for activation of the circadian systems of nocturnal rodents commonly used as cancer models can be as much as 10,000 times lower than the threshold levels needed to activate the circadian systems of diurnal species such as humans. As suggested above, this profound species difference is clearly demonstrated in the work of Blask and his colleagues, where only 0.1 μW/cm² of white light was sufficient to reliably suppress nocturnal melatonin in rats 2033, but 580 μW/cm² (2800 lux) was necessary to suppress nocturnal melatonin in women by 40% 20. Not only is this level several orders of magnitude greater than the threshold for melatonin suppression in nocturnal rodents, it is equivalent to daytime light levels found outdoors during the morning and not at all similar to light levels produced by architectural LAN, especially in residences (Table 1). Further, experimental data suggest that nocturnal rodents such as mice 34 do not reliably exhibit spectral opponency in circadian phototransduction, although humans and diurnal primates do appear to exhibit opponency 35363738. These comparisons among species underscore the importance of properly characterizing light as a stimulus to the circadian systems in both humans and in the nocturnal rodents used as animal models for understanding human breast cancer etiology.

Light levels in animal laboratories designed in the 1990s have typically been specified to be about 600 to 800 lux of white light on a horizontal plane one meter above the floor 39, corresponding to an irradiance of about 150 μW/cm². More recently, recommendations 40 have been reduced to about 300 to 325 lux (about 70 μW/cm²) to help reduce the incidence of retinal damage in nocturnal rodents by lighting. Regardless, these light levels would still be likely to result in illuminances at the eyes of rodents housed in such laboratories well above the 0.1 μW/cm² threshold for melatonin suppression 2033. Such disparities between typical human light exposures and the conditions used in the study of cancer in nocturnal rodents will need to be much better addressed before a quantitative understanding of architectural LAN and breast cancer risk can be developed from laboratory studies where animal models are used.

Once quantitative estimates of the light levels experienced by women at night have been determined, either for interpreting epidemiological studies or for designing parametric laboratory studies using animal models, the functional relationships between complex light exposures and melatonin suppression in humans must then be established. These relationships can then be meaningfully used to begin to assess the importance of LAN for suppressing nocturnal melatonin in actual architectural spaces. A model of human circadian phototransduction relating light exposure in laboratory environments to nocturnal melatonin suppression was developed by Rea and colleages 37. Figure 3 is based upon that model and shows nocturnal melatonin suppression in humans as a function of the modeled circadian light stimulus. The solid line is the best fitting functional relationship between the modeled circadian light stimulus and nocturnal melatonin suppression values for humans from a number of empirical studies using both monochromatic and polychromatic light 354142434445. The circadian light stimulus values on the abscissa in Figure 3 are for a 60 minute light exposure, but, obviously, durations shorter or longer than 60 minutes will have less or more effect on melatonin suppression for the same circadian light level and spectrum 41. In fact, the solid line in Figure 3 is based on data from 30 43, 60 35414445, and 90 minutes 42; the data were adjusted empirically to account for those different durations. Also shown in this figure are the resulting circadian light stimulus values for varying illuminances of two conventional light sources, incandescent and daylight illuminants 46.

Figure 4 shows the individual curves of the same form as in Figure 3, fitted to the 30-, 60- and 90-minute data separately. Table 2 shows the predicted values of nocturnal melatonin suppression for incandescent and daylight illumination, at different light levels and for different exposure durations; the values are based upon the three functions in Figure 4.

It should be noted that the 90-minute suppression values for the daylight source are consistent with those recently published by Zeitzer and colleagues 47, who found a half-maximal value of melatonin suppression in subjects exposed to three consecutive (over three days) 5-hour pulses of light at an illuminance (at the eyes) of around 100 lux. The uncertainty in the predicted values in Table 2 is, however, reflected in Figure 3. Empirically, for a specific circadian light stimulus value, the range in mean nocturnal melatonin suppression would be expected to be approximately 20%, or ± 10% of the predicted value. This level of uncertainty is also consistent with the confidence intervals for predicted melatonin suppression identified by Zeitzer and colleagues 47.

The studies of human nocturnal melatonin suppression to light in the laboratory illustrated in Figure 3 show that a reliable threshold for measurable melatonin suppression (approximately 15%, given the approximately ± 10% uncertainty in prediction of melatonin suppression) is, depending upon the spectral power distribution 37, about 30 lux of white light at the cornea for a 60-minute exposure 414445. Based then on the values in Table 2, it can be inferred that melatonin levels should be largely unaffected by brief (i.e., less than 30 minutes) exposures to the relatively low ambient levels of white (incandescent) light typically found in residences (i.e., less than 30 lux). Threshold values will undoubtedly be different for different individuals with different light histories 293031 and with different demographic characteristics 4849, but 30 lux for 30 minutes is proposed here as a working expected threshold value for women living in North American residences until more specific data are available. It is important to note again, however, that when additional task lights are used in the home for critical visual tasks, such as reading, sewing or grooming, light levels at the cornea can reach 150 to 200 lux at the eye 26. Working under these light levels for several hours in the evening should in fact impact the onset of nocturnal melatonin production, but not necessarily in a clear or unambiguous way. Two studies have shown that LAN experienced by diurnally active adults in their homes has no impact on total nocturnal melatonin production, and only a small effect on phase 5051. On the other hand, in long sleepers (sleep duration longer than 9 hours) the duration of nocturnal melatonin production was greater than that in short sleepers (sleep duration less than 6 hours) under a constant routine protocol and exposure to less than 10 lux at the eye, although the amplitude of melatonin production was not significantly different 52. Wehr 53 showed that the duration of nocturnal melatonin production was longer when subjects were exposed to a short photoperiod compared to a long photoperiod. Unfortunately, however, the quantity and spectrum of the light exposure was undefined, both indoors and outdoors. In sum, although it seems like long or short sleepers will differ in their nocturnal melatonin production duration, no studies have investigated whether this is due to individual differences (i.e., one is a long sleeper because he/she naturally produces melatonin for a longer duration over the course of the night) or if LAN typically found in real-life applications affects melatonin production duration in either long or short duration sleepers.

It is also important to stress that although a working threshold dose of light for nocturnal melatonin suppression in the home (i.e. 30 lux of white light for 30 minutes) could possibly be a useful starting point for discussion of a "safe" amount of light at night, it is clearly beyond our collective current understanding to suggest that this value, or any other, is a "safe" dose of light. It is equally premature to suggest that LAN is "unsafe," particularly without a proper, species-specific quantitative framework for such statements. What will be meaningful in the context of women's health is to base such discussions on existing data, or to generate new data, and to begin to move from a qualitative to a quantitative discussion of circadian LAN.

In the context of safety and health, it is also important to reconsider one purpose of LAN, namely to provide people with a "safe" nocturnal environment via properly specified emergency lighting 5455 or security lighting 5657. Arguably it is safer to use light of modest intensity and duration to reduce risks of trips and falls during nocturnal visits to the bathroom and kitchen than to leave the lights off or to use lighting that is insufficient for visual navigation at home. In this context, however, safe navigation throughout the house does not require high light levels 55. For example, vanity lights around a mirror operated by a switch located at the bathroom door usually provide much higher light levels at the eye than are necessary for safe navigation 26. For safe movement through the house in the middle of the night it is usually only necessary to employ a secondary, low-level lighting system that does not exceed 30 lux of white light at the cornea for more than a few minutes. Some have suggested that only "red" dim night-lights be used, but this recommendation seems unnecessarily restrictive, especially because the visual system is not particularly sensitive to long-wavelength radiation and, therefore, safe navigation could be seriously hindered. In fact, utilizing light levels high enough for safe navigation but low enough to minimize circadian stimulation for short exposure times (less than 30 minutes) is especially important for senior adults who can have difficulty seeing and walking 58. Falls among older adults have been associated with increased mortality 59. To avoid risk of trips and falls due to temporary loss of sensitivity during dark adaptation after lights are shut off, a secondary, low-level lighting system for nighttime applications could, again, be quite important to health and well-being of senior adults 60.

It is also worth mentioning concerns expressed about the possible impact of light from the street or from adjacent properties entering the bedroom at night after lights in a residence have been turned off. These light levels rarely exceed 10 lux at the cornea outdoors (Table 1). Indoors, behind closed curtains, the levels would be likely to be much lower. Further, the human eyelids transmit only about 1% to 3% in the short wavelength region of the visible spectrum 61, so 1000 to 3000 lux at the closed eyelid would be needed for one hour to reach the threshold for melatonin suppression in humans. In fact, Jean-Louis and colleagues 62 showed that a one-hour exposure of 1700 lux of white light at the eyelids (delivered via a light mask) did not suppress nocturnal melatonin in humans. Hatonen and colleagues 63 showed that, on average, 2000 lux of white light applied to closed eyelids for 1 hour did not significantly suppress melatonin, although a quarter of their subjects did show some suppression. In this regard, individual differences may become a more important consideration in the future. Nevertheless, their conclusion was that indoor light was likely ineffective to the circadian system while one is asleep because light levels typically found in indoor environments are much lower than 2000 lux at the eye. Arguments for using black-out shades to limit light in the home at night because of its possible link to cancer risk have no quantitative foundation for melatonin suppression within the context of people living normal lives and working day-shifts. Of course, light trespass in our bedrooms should be avoided, but for the right reasons: light trespass is annoying and wasteful. Given the available published data on human melatonin suppression in response to light, light trespass through residential windows is an unlikely cause of melatonin suppression, simply because the light levels are so low, particularly with the eyes closed.

Obviously, light is required to perform night-shift work. Because LAN can suppress nocturnal melatonin and because no instruments have previously been available to conveniently and accurately measure circadian light exposure, epidemiological studies have used night-shift work as one surrogate for LAN. In fact, studies on the relationship between LAN actually experienced by night-shift women and melatonin depletion have not been consistent with the assumption that LAN always suppresses melatonin 6465. Nevertheless, there should be a real societal concern about women performing night-shift work. As previously noted, a growing number of epidemiological studies show that night-shift work places women at a higher risk for breast cancer 91011.

Measurements in several office and hospital workplaces (Table 127) clearly show that vertical illuminances from white light likely to be experienced in the workplace for durations longer than 30 minutes can regularly exceed 30 lux. Measurements of light levels in neonatal intensive care units 66, for example, operated both day and night also showed that light levels could be high enough to result in nocturnal melatonin suppression (Table 2). It is reasonable to suppose then that night-shift women experience light levels bright enough and long enough to suppress nocturnal melatonin while they work at night. The data in Figure 2 suggest, if these women are typical, that the light levels to which they are exposed during the day are lower than those experienced at night, so that night-shift nurses do experience a regular, albeit muted and perhaps irregular, light-dark cycle. What is not yet clear, of course, is the circadian phase of these women with respect to their light-dark cycle. Data from Burgess and Eastman 5051 suggest that the total melatonin levels across the 24-hour day might be unaffected by shortened sleeping schedules even though the phase of melatonin production is altered by both evening and morning light and by the sleep-wake cycle. In the context of night-shift work, little is known about the impact of long-term adaptation to different light-dark cycles over time 2930. LAN should not then be considered out of context of the light-dark cycle actually experienced by night-shift women and without a much better understanding of workers' adaptation to their working and sleeping schedule.

Night-shift women may also be at a higher risk of tumor growth for reasons apart from those associated with LAN. There are potential dietary differences between day- and night-shift workers 67 and this could be especially important if night-shift workers are consuming higher levels of linoleic acid 7. Caffeine (100 mg) given four times a night was shown to reduce onset of melatonin levels in women in their luteal phase in the same manner as bright light 68. Studies have also shown that sleep deprivation negatively impacts the immune system 6970 and night-shift workers are generally sleep deprived because the quality of daytime sleep is often worse than nighttime sleep. It is important to add, however, that a recent epidemiological study failed to statistically show that occupational factors, such as sleep rhythms disruptions, were positively related to breast cancer risk among Finnish flight attendants 71. Schernhammer and colleagues 72 also examined the relationship between melatonin and lifestyle factors and other endogenous sex steroid hormones. They concluded that age, body mass index, and heavy smoking were significantly related to lower levels of melatonin, while parity was significantly related to higher levels of melatonin. No other reproductive factors or sex steroid hormones were significantly associated with melatonin levels. Of course, more studies are needed to confirm their results. Moreover, although melatonin has been shown to have oncostatic effects (for review, see 73), it may not be the only hormone important to breast cancer risk 7475. Clearly then, the higher risk of breast cancer associated with night-shift work is not related to LAN in a simple, easily defined way. Indeed, LAN associated with night-shift work may be entirely beneficial to night shift workers in helping entrain their circadian rhythms to the nighttime hours, and very different than LAN associated with diurnal lifestyles.