List of reviewed studies, organized chronologically by publication year1

REF2YearStudy type3Cohort size and study designMain resultsLimitations
ALAN
  • ALAN and sleep

[111]2002Field154 postmenopausal women, with a mean age of 66.7 years, underwent one-week continuous recordings of light exposure and sleep using wrist actigraphy. Mood was evaluated using a questionnaire.Pre-bedtime illumination (median 24 lux) had no significant impact on sleep. However, total 24-hour light exposure correlated with shorter sleep latency (P < 0.01), wakefulness within sleep (P < 0.05), and depressed mood (P < 0.01)A homogeneous cohort in age and gender may influence light preferences and health. Actigraphy sensors may not fully capture light exposure, and ambient light recordings are limited. Spectral composition was also not considered
[10]2013Lab10 healthy volunteers were exposed to two conditions over two nights: no light and 40 lux luminance, while PSG monitored sleep parameters.Nights with lights-on were associated with increased N1 sleep, decreased SWS, and increased arousal. Moreover, brain activity was observed to decrease during NREM sleep under lights-on conditions.Small cohort sizes and controlled conditions may not fully reflect real-world scenarios. The study did not account for spectral composition and relied solely on a single light intensity.
[114]2013Lab30 individuals were exposed to light sources with temperatures of 6,500 K, 2,500 K, and 3,000 K at intensities ranging from 27 to 29.2 lux. The study protocol included dark conditions followed by 2 h of exposure sleep metrics were monitored using a PSG device.All-night NREM EEG power density showed lower frontal power with 6,500 K light compared to 2,500 K (0.8%) and 3,000 K (0.2%). Additionally, SWS activity decreased by 0.4% and 0.3%, and total sleep time was shorter by 4 min and 12.7 min with 6,500 K light compared to 2,500 K and 3,000 K, respectively.A small cohort size and controlled conditions may not accurately reflect real-world scenarios. Additionally, the study did not account for changes in light intensity.
[110]2017Field20 healthy volunteers were monitored for three weeks using PSG and actigraphy to collect sleep metrics. Their light exposure was monitored using wrist actigraphy.Later circadian phases were linked to lower light intensity (P < 0.01), delayed light exposure (P < 0.001), and later sleep timing (P < 0.001). Light exposure above 10 lux increased awakenings (P < 0.05), while delayed exposure reduced REM latency (P < 0.05). Lower light levels were associated with more REM sleep (P < 0.001), and higher light intensity with early exposure led to greater SWS accumulation (P < 0.05).Limitations include a small cohort, potential obstruction of actigraphy light sensors, and reliance on ambient light measurements that may not accurately reflect perceived light. Additionally, the study did not account for the light’s spectral composition.
[115]2017Lab19 volunteers (mean age 24.3 years) were exposed to light from computer screens with varying intensities (80 lux and 350 lux) and wavelengths (460 nm and 650 nm). Sleep metrics were collected with PSG, MLT levels from urine, and body temperature were recorded at set times. Behavioral evaluation was performed using questionnaires.SWL exposure was linked to shorter sleep duration (8.6 min to 16.4 min), increased WASO by 2.1% to 3.1% (P < 0.01), decreased SE by 2.1% to 3.1% (P < 0.01), and reduced SWS by 6.5% to 7.7% (P < 0.05). On SWL exposure nights, body temperature was higher, MLT levels were lower (P < 0.05), and subjective sleep quality, mood, and attention were worse, with attention impairment significant (P < 0.05).The 22-inch LED screens used may not represent other technologies or sizes. Measurements were limited to one night per light condition, restricting long-term effects. Daytime exposure, which influences MEL levels and circadian function, was not recorded. The small cohort and controlled setting may not reflect real-world scenarios.
[112]2023Field59 volunteers wore wrist actigraphy to measure light exposure and collect sleep metrics over seven days, from February to July 2022. Personal attributes were collected using questionnaires and sleep diariesEarlier bedtimes were linked to afternoon light exposure (13:00–17:00). Longer bedtimes corresponded to extended sleep on high light exposure (> 1 lux) days, while more natural light was associated with earlier sleep onset. ALAN exposure above 10 lux before sleep was tied to earlier sleep onset, peaking in the last 30 min before bedtime. Each 1 log lux increase in light exposure extended sleep onset by 0.5 h (P < 0.05).The five-month data collection period may introduce seasonal bias. Questionnaires are prone to omission and recall biases. Actigraphy light sensors could be obstructed, recording ambient rather than perceived light. Additionally, the spectral composition was not considered
  • ALAN and MLT

[119]1980Lab 6 volunteers underwent a two-night protocol: 500 lux on the first night and 2,500 lux on the second. Two volunteers continued for two additional nights, with 1,500 lux on the third night and darkness on the fourth. Light intensities were measured at eye level, and MLT synthesis was assessed via blood samples.MLT levels decreased (P < 0.05) within 10 min to 20 min of exposure to 2,500 lux incandescent light, nearing daytime levels within 1 h (P < 0.01). MLT levels quickly rebounded to pre-exposure levels within 40 min after returning to darkness. Exposure to 500 lux did not affect MLT levels significantly. Subjects exposed to 1,500 lux light showed intermediate MLT levels, with recovery similar to that after exposure to 2,500 lux.The small cohort size and controlled conditions may not accurately reflect real-world scenarios. Additionally, the study did not account for the spectral composition or personal confounders of the volunteers.
[124]1995Lab 11 blind patients with no conscious perception of light and 6 healthy men aged 20 years to 25 years were exposed to 10 lux to 15 lux during baseline and 6,000 lux to 13,700 lux during stimulus exposures. Body temperature and blood samples for plasma MLT evaluation were collected at fixed intervalsAll healthy individuals showed MLT suppression in response to bright light, with levels averaging 66% lower during the last 60 min of exposure compared to 24 h earlier. Blind patients who did not experience sleep difficulties, also showed MLT suppression, with a 69% decrease following bright light exposure. These blind patients maintained stable entrainment to a 24-hour day in repeated measurementsA small cohort size and controlled conditions, which may not fully represent real-world scenarios.
[120]2001Lab15 volunteers (mean age 31.8 years) were exposed to light spectra (470 nm to 660 nm) aimed at the pupil. Over a 6-week protocol, they experienced various spectral conditions. Saliva MLT was sampled at fixed intervals.The shorter wavelengths (470 nm, 497 nm, and 525 nm) caused significant MLT suppression, ranging from 65% to 81% (P < 0.0001). They also resulted in the greatest delay in simulated light MLT onset on the second night, ranging from 27 min to 36 min.A small cohort size and controlled conditions may not accurately represent real-world scenarios.
[12]2005Lab48 individuals, with a mean age of 23 years, were exposed to light intensities ranging from < 0.03 lux to 9,500 lux over three consecutive days. During the study, body core temperature and blood plasma MLT levels were monitored at regular intervals.Significant shifts in MLT peak midpoint were seen at all light intensities compared to < 0.03 lux (P < 0.05). Suppression was minimal below 15 lux but complete above 200 lux. The circadian system responds logarithmically, with half-maximal effects at 50–160 lux and saturation at 1,400 lux. MLT suppression reached half-max effects at 10–200 lux and saturation at 1,100 lux.Background light could significantly influence the responsiveness of the circadian timing system. Additionally, the study did not account for spectral composition. Lastly, the study was conducted under controlled conditions, which may not accurately reflect real-world scenarios.
[125]2007Lab2 blind subjects participated in parallel experiments: a 56-year-old male and an 87-year-old female, both without light perception. In Experiment #1, the male underwent a 14-day inpatient study with random exposure to 555 nm and 460 nm light for 6.5 h, with plasma MLT measured via blood samples and sleep assessed by EEG. In Experiment #2, the female’s pupil constriction response to various monochromatic light wavelengths and irradiances was tested for 10 s.In Experiment #1, 555 nm light had no effect on plasma MLT, while 460 nm light reduced it by 57% and caused a 1.2-hour phase delay in the circadian MLT rhythm, compared to a 0.4-hour delay with 555 nm light. Blue light increased alpha EEG activity and reduced subjective sleepiness, consistent with its effects in sighted individuals.
In Experiment #2, irradiance-response curves matched vitamin A opsin-pigment nomograms (P < 0.01) for all three cones, indicating pupil constriction was driven by a photopigment with a peak at 476 nm. This suggests that ipRGCs mediate both pupillary and circadian responses to light, even in the absence of rods and cones.
A small cohort size and controlled conditions may not fully represent real-world scenarios. Additionally, the male subject received a pupil dilator before each experimental light exposure. While this approach is acceptable in experimental settings, it may not fully replicate real-world conditions of light exposure.
[129]2011Retro116 healthy volunteers aged 18 to 30 participated in two inpatient studies. In Study #1, 104 volunteers followed a 9–10 day protocol involving enforced wakefulness for 30–50 h. In Study #2, 12 volunteers were exposed to 200 lux daily for 14 days, followed by a 40-hour constant routine under the same light intensity. Blood samples were collected at intervals to monitor plasma MLT concentrations.In Study #1, MLT onset occurred 23 min before sleep under room light conditions (< 200 lux) but 1 h and 57 min before bedtime under dim light (P < 0.05), marking a 94-minute earlier onset. Room light reduced MLT secretion duration by 1 h and 32 min compared to dim light (P < 0.05). Additionally, 78.6% of volunteers had earlier MLT onset by over an hour, and MLT concentrations were reduced by 71.4% on average under room light. In Study #2, four of five individuals showed significant MLT suppression with room light exposure, while seven of eight experienced greater suppression with higher light intensities (~200 lux).Study conditions, specifically dim and bright light exposures, were not balanced, which could affect the accuracy of comparisons. Additionally, the study did not account for spectral composition. Lastly, the studies were conducted under controlled conditions that may not fully reflect real-life scenarios.
[122]2013Lab 9 volunteers (mean age 26.3 years) were exposed to six light conditions, varying in intensity from < 10 lux to 500 lux and temperature from 2,000 K to 6,000 K, over six days. MLT concentration was measured from saliva, and alertness was assessed via questionnaires.Comparisons showed significant decreases in MLT levels with three of the four lamps containing blue components (P < 0.03). The 500 lux, 6,000 K condition did not significantly affect MLT levels (P = 0.07), and the 130 lux, 2,000 K condition had no effect. Subjective alertness significantly increased after exposure to the 130 lux, 6,000 K; 500 lux, 6,000 K; and 500 lux, 5,000 K conditions (P < 0.05).The study’s small cohort size and controlled conditions may not fully represent real-world scenarios. Additionally, as suggested by the authors, it lacks outcome variables such as sleep parameters and objective measures of behavioral variables.
[126]2018Lab18 healthy blind subjects were exposed to 6,807 lux from a 4,100 K source. The study included 5–6 days of inpatient observation followed by a 38-day protocol where 10 volunteers adjusted to a 28-hour “day” which included light stimuli. Saliva and blood samples were collected to measure MLT concentration, and body core temperature was recorded at set intervals.During 6.5 h of light exposure, MLT concentrations decreased by an average of 71.6% in five blind participants, with each showing reductions over 33%. Four had a normally phased 24-hour rhythm in urinary MLT before the study, while the fifth, with an irregular rhythm and opaque scleral shells, showed a positive response after removing the shells. Among those without MLT suppression, only one had a 24-hour urinary rhythm prior to the study, likely due to non-photic cue entrainment.A small cohort size and controlled conditions, which may not fully represent real-world scenarios.
[121]2019RetroTo model the relationship between alpha-opic illuminances and MLT suppression, early study results were reanalyzed and validated.The study found the highest R² value (0.87) with melanopic lux and melanopsin-weighted irradiances, indicating that MLT suppression is strongly predicted by melanopsin-weighted illuminance. MLT suppression was found to be most sensitive at 480 nm SWS light and is enhanced by blue-enriched light stimuli. Moreover, the model demonstrated good predictive performance.The model is based on a single dataset capturing a narrow circadian phase, limiting its applicability to other phases and conditions. Moreover, its reliance on controlled settings may not reflect real-world scenarios.
[123]2022Lab 100 healthy volunteers (aged 18 to 30) completed a nine-day protocol involving 30- and 50-hour constant routines and exposure to various light spectra and intensities. Saliva and blood samples were collected to assess MLT secretion.MLT suppression and circadian phase resetting were best modeled with peak sensitivities at 481 nm and 483 nm, respectively. Initially, MLT suppression peaked at 441 nm and 550 nm, indicating significant early contributions from S and L+M cones. Over time, the peak shifted to 485 nm, reflecting a predominant influence of melanopsin. Similarly, circadian phase resetting showed peak sensitivities from 445 nm to 487 nm, with a shift from S-cones to melanopsin as exposure continued.The study was conducted under controlled conditions, and participants used a pupil dilator, which may not reflect real-life scenarios. Additionally, it was not documented whether maximum pupil dilation was maintained throughout the 6.5-hour light exposure, potentially affecting the results.
[127]2022Field580 participants (mean age 71) monitored short-wavelength ALAN exposure and sleep metrics over two days using a headband for light, actigraphy for sleep data, and a PSQI questionnaire for subjective evaluation. Daily urine samples assessed MLT levels. The study was conducted from November 2012 to March 2014.Increased short-wavelength ALAN was significantly linked to higher global PSQI scores and sleep disturbances (P = 0.004 and P = 0.006). Multivariable analysis showed a higher OR for disturbances in the highest tertile compared to the lowest (P = 0.006). MLT levels were significantly lower in the highest tertile (P = 0.039). Objective sleep measures revealed longer sleep onset latency (P < 0.001), increased WASO (P < 0.05), and decreased SE (P < 0.001) in the highest tertile.The study’s extended duration raises concerns about seasonal effects. Despite a large cohort, each volunteer was observed for only two days, limiting individual data. Inconsistent MLT and sleep measurements, a cross-sectional design, and reliance on a homogeneous elderly population limit generalizability and introduce potential bias.
  • Health effects of ALAN exposure and MLT suppression

[132]2007Lab10 volunteers (mean age 21.9) were asked to wear wrist actigraphy for one week to measure light exposure and to sleep in the lab during a night session with 2 h of exposure to 1,000 lux bright light. Saliva samples were collected for MLT measurement, and subjective sleep data were obtained via questionnaires. The study, conducted in winter and summer, to assess seasonal effectsAwakening occurred about 30 min earlier in winter (P < 0.05), with higher MLT amplitude. MLT concentration decreased with light exposure, but the suppression percentage was significantly greater in winter (66.6%) compared to summer (37.2%) (P < 0.01). Winter ambient light levels were about half those of summer (P < 0.01). There was considerable variability in MLT levels, light exposure, and suppression among individuals. Other sleep and MLT metrics showed no significant seasonal differences.The small cohort size and controlled conditions may not accurately reflect real-world scenarios. Additionally, using a wrist-worn device to measure light could be inaccurate if the light sensor becomes obstructed. Moreover, the study did not account for variations in light spectra, which could differ between ALAN sources and sunlight across different seasons.
[157]2019Lab 11 healthy volunteers (mean age 23 for the control group, 27 for the exposure group) underwent a 4-day simulated night shift protocol. After a 24-hour baseline, participants followed a 10-hour delayed sleep schedule. The control group was exposed to dim light (< 10 lux), while the exposure group received 6,500 lux of bright light. Blood samples were for MLT measurement and microarray analysis.In a study of 492 probe sets with a 24-hour rhythm, 67% showed altered circadian expression under bright light during a simulated night shift, with gene rhythms delayed by 7.8 h on average. The bright light group had a significantly greater delay in the midpoint of MLT secretion (7.6 h) compared to controls (0.4 h, P = 0.014). Gene analysis indicated bright light's strong phase-shifting effect, affecting genes related to immune processes, lipid metabolism, and circadian rhythm regulation. The study is limited by a small cohort size, a narrow age range (18–30 years), and a skewed sex distribution (10 males vs. 1 female). Furthermore, the controlled conditions employed may not fully replicate real-world scenarios.
Noise
  • Noise sleep and CORT

[193]1983Field26 healthy individuals, who had lived near a busy street slept under various conditions, including changes in bedroom location, earplugs, and double-glazed windows, to simulate both quiet and typical nighttime environments. Their sleep behaviors were monitored using EEG signals, while sound was recorded with microphones and tape recorders.Noise peaks of 50.3 dB(A) can cause awakenings, while changes in sleep state occur at 48.5 dB(A), and transient reactions at 47.6 dB(A). The study suggests that limiting noise peaks to below 40 dB(A) can prevent 80% of sleep disturbances and 87% of awakenings while keeping noise below 45 dB(A) can avoid two-thirds of sleep impairments. Additionally, peak noises during quiet conditions caused shifts in sleep at lower intensities [42–44 dB(A)] compared to noisier conditions [50–53 dB(A)].Changes in sleep settings affected room temperature and other environmental factors, which may have influenced sleep. This complicates the observed relationship between noise and sleep, as other interactions could have played a role. Additionally, the study’s small sample size limits the generalizability of its findings.
[194]1987LabThe study included a pilot with one participant exposed to noise levels from 40 dB(A) to 70 dB(A) in 5 dB(A) increments, and a main study with 9 individuals exposed to: (a) a single truck pass, (b) continuous all-night traffic noise, and (c) a combination of both. Sleep metrics were recorded using PSG, and mood was assessed via questionnaires.Awakening reactions in response to car passages were observed at 55 dB(A) (P < 0.05). An increase in noise level from 45 dB(A) to 55 dB(A) significantly affected sleep stages (P < 0.01). Additionally, noise peaks relative to background levels were better predictors of arousal and changes in sleep stages than absolute noise peak levels. Subjective mood and sleep quality questionnaires supported these findings.The study’s small cohort size and controlled conditions may not fully reflect real-world scenarios.
[195]1990Lab 18 individuals (mean age 20.8) were divided into three groups. The first and second groups were exposed to infrasound and low-frequency sounds (10–63 Hz) at 50–105 dB(A). The third group was subjected to simulated traffic noise (25–1,600 Hz) at 40 dB(A), 50 dB(A), and 60 dB(A) for 30 s every 20 min during sleep. Sleep patterns were monitored using EEG.The reaction rate during N1 was significantly higher than in other sleep stages, with REM and N2 being more affected than N3 (P < 0.01), indicating that lighter sleep stages are more sensitive to infrasound and low-frequency sounds. However, in groups one and two, the relationship between infrasound, low-frequency sounds, and sleep metrics was insignificant compared to control nights. In group three, exposure to 40 dB(A), 50 dB(A), and 60 dB(A) led to awakening rates of 33%, 50%, and 100%, respectively, with 40 dB(A) also causing changes in sleep stages in 51% of cases.The study’s small cohort and controlled conditions may not reflect real-world scenarios. It also did not account for indoor noise from appliances, which could impact results. Additionally, variability in noise perception among the three groups may introduce bias and affect the findings.
[191]2001RetroThis study is based on the results of road traffic noise and urine catecholamine levels during sleep. Overnight urine samples were analyzed for adrenaline and NA. Noise exposure was assessed through traffic counts, and indoor exposure was evaluated through questionnaires.A significant increase in renal adrenaline was associated with a higher number of cars per day (P < 0.05). Additionally, subjective perceptions of indoor noise by volunteers were strongly correlated with urinary NA secretion (P < 0.001). Renal adrenaline levels were 26% higher in women experiencing high levels of disturbance compared to those with low disturbance (P < 0.05).The study lacked data on daytime noise exposure, making it difficult to assess cumulative effects on adrenaline and NA secretion. Additionally, noise exposure was estimated from traffic counts rather than direct measurements, potentially misrepresenting actual noise conditions.
[192]2002Field16 volunteers were exposed to airport noise for 40 nights, peaking at 65 dB(A). The noise occurred from 23:00 to 06:00, with an average continuous level of 42 dB(A). Urine samples were collected nightly to measure CORT, adrenaline, and NA, and subjective data were gathered via a questionnaire.Nocturnal noise exposure significantly increased CORT levels in men but not in women, suggesting greater men sensitivity to noise (P = 0.038). Variability in adaptation types was observed, with changes in CORT excretion patterns and significant associations between adaptation differences and CORT variations (P = 0.004).The study is limited by its small sample size and its exclusive focus on airplane noise, which restricts the generalizability of the findings to other common noise sources, such as road traffic and indoor appliances.
[206]2003Lab 10 volunteers (mean age 24.7) were exposed to 8 h of 75 dB(A) mill noise. Sleep was monitored with PSG and questionnaires, and blood samples were collected to measure CORT levels.On the exposure night, both REM sleep and total sleep durations were significantly reduced compared to quiet conditions (P < 0.001 and P < 0.05, respectively). REM sleep onset was delayed (P < 0.01), and there were more frequent stage shifts (P < 0.001). SE fell below 80%, and CORT levels were elevated during the exposure night.The controlled conditions and small sample size limit the study’s applicability and generalizability to real-world settings. Additionally, the brief exposure duration restricts the assessment of continuous effects.
[209]2006Field360 individuals, equally divided between children (mean age 10.9 years) and parents (mean age 42 years), provided sleep habit information through questionnaires and interviews. Noise exposure was categorized by road traffic levels [< 55 dB(A), 55–59 dB(A), 60–64 dB(A), and > 64 dB(A)].For parents, road traffic noise significantly (P < 0.05) affected sleep quality, awakenings, and perceived interference. Children also showed a notable link between noise and sleep quality, as well as daytime sleepiness. Despite reporting better sleep quality and fewer awakenings, children’s wrist-actigraphy showed that parents had better objective sleep (P < 0.001). Both groups experienced acute sleep disturbances at 60 dB(A) to 64 dB(A) noise levels.Actigraphy, which tracks body movements, may misclassify children’s sleep due to their higher sleep activity compared to adults. The study’s noise assessment, categorized by road traffic noise, might not accurately reflect direct exposure. Additionally, sleep questionnaires are prone to recall and response biases.
[72]2011Lab12 volunteers with a mean age of 27 years underwent a 3-day lab exposure protocol, with the first night serving as an adaptation period. On the second and third nights, noise stimuli were presented every 30 s during sleep, starting at 40 dB(A) and increasing in 5 dB(A) increments until arousal, a change in sleep stage, or a maximum of 70 dB(A). PSG was used to collect sleep metrics.Increased sound levels significantly disturbed sleep, with effects varying by sound type. Arousal probabilities differed among sleep stages: N2 was different from N3 and REM (P < 0.001), while N3 and REM were similar. Arousal rates were lower during N3 compared to N2, and REM had a more consistent arousal response to different sounds. Sound-induced arousals significantly increased HR, with the highest responses during REM, followed by N3 and N2. Additionally, HR response times from arousal onset to peak were shorter during REM sleep.The small cohort, narrow age range, and controlled conditions limit the study’s generalizability. Real-world noise exposure is more variable and interactive, which may amplify physiological and psychological effects differently than the fixed noise administration used in the study.
[196]2011Field40 healthy volunteers, with a mean age of 48.5 years, participated in the study. Half of the cohort was exposed to road traffic noise, while the other half was exposed to rail noise. Data were collected over two consecutive nights, including sound recordings and PSG to measure sleep metrics.Road traffic noise did not affect sleep, but rail noise exposure was significantly linked to increased WASO and reduced REM sleep (P < 0.01). A significant difference was found in arousals and REM sleep between rail noise levels below and above 50 dB(A) (P = 0.02). Specifically, exposure to rail noise above 50 dB(A) resulted in shorter REM sleep, while exposure below 50 dB(A) led to more arousal.The study did not control for non-traffic noises, and unattended data recording prevented sensitivity adjustments. Personal confounders and individual noise sensitivity were not analyzed. The small sample size and brief duration may not fully capture noise effects or variability, limiting generalizability and may introduce bias.
[198]2012RetroData from 7,019 participants (mean age 50.5) in the Finnish Public Sector Study Cohort (2004–2009) were analyzed for sleep duration and insomnia. Nighttime traffic noise levels were calculated, and personal confounders and socioeconomic attributes were assessed via questionnairesParticipants exposed to noise levels > 55 dB(A) had significantly higher OR for insomnia symptoms and nonrestorative sleep compared to those exposed to ≤ 45 dB(A). Frequent nighttime awakenings were notably linked to higher anxiety scores at > 55 dB(A). The impact of noise exposure > 55 dB(A) on insomnia was more pronounced among women and obese individuals compared to men and those who were not obese.The study’s generalizability is limited by its homogeneous population and incomplete traffic data. Key limitations include exposure misclassification, missing nighttime traffic intensity data, and potential bias from variations in noise levels and survey timing.
[201]2014Lab 103 patients, with a mean age of 60 years, across 29 ward rooms, completed the PSQI and the Leeds Sleep Evaluation Questionnaire to evaluate their sleep performance. Noise exposure during one night of sleep was measured using type-1 noise level dosimeters.The study found significant declines in sleep quality among patients. Overall, 86% reported disturbed sleep, linked to daily average noise levels of 63.5 dB(A) (P < 0.05). Female patients had higher PSQI scores (P = 0.06), and those with severe disease had poorer sleep (P = 0.04). Patients not on sleep-interfering medications fared better (P = 0.08). Noise levels above 49.3 dB(A) during the day and 34.2 dB(A) at night were associated with increased disturbances, with common sources being caregivers, visitors, other patients, toilet flushing, and medical devices.The small cohort limits statistical power and generalizability. Sleep assessments via questionnaires may suffer from omissions and recall bias. Dosimeters did not capture detailed noise characteristics, and psychological factors like stress and depression, which are common among patients, were not considered.
[210]2015RetroThe study, based on a 2007 environmental and health survey in Sweden, included 2,612 participants (median age 46 years) living near rail and noisy roads. Noise exposure was categorized into six levels: < 40 dB(A), 40–44 dB(A), 45–49 dB(A), 50–54 dB(A), 55–59 dB(A), and ≥ 60 dB(A). The survey assessed participants’ annoyance, sleep quality, bedroom window orientation, and subjective noise perception.Access to a quiet or green space decreased with higher noise levels. A bedroom window facing green space significantly reduced noise annoyance (P < 0.05) and concentration problems. Higher noise exposure was linked to poorer sleep quality. Men reported fewer sleep issues compared to women, while those under financial strain experienced more sleep problems.Limitations include reliance on potentially biased self-reported data, unconsidered factors like OSA, and unvalidated scales. With a 54% response rate, there may be selection bias, especially in high-noise areas. Noise assessments lacked details on building characteristics and other sources, and combining noise types could skew results. Assigning noise exposure by building rather than individual units may also introduce errors.
[203]2018Lab 64 patients, averaging 63.9 years, completed RCSQ. Noise exposure was recorded for one night using a microphone and laptop.A daily average sound pressure of 54 dB(A) was associated with poorer sleep (P < 0.05). Additionally, women reported better sleep quality than men (P < 0.01).Small cohort limits generalizability. Sleep questionnaires may introduce recall and response biases, especially in critically ill patients. A single night’s measurement does not capture long-term effects.
[207]2018Lab 48 day-shift workers (mean age 45.1) were split into two groups: one experienced high noise followed by low noise, and the other the reverse. CORT levels were measured from blood samples, and sleep was recorded using PSG.Sleeping after exposure to 76.8 dB(A) was linked to longer sleep durations but decreased SWS and SE (P < 0.05). Additionally, blood pressure, sympathetic activity, and CORT levels significantly increased following noise exposure (P < 0.05).Controlled conditions and a small sample size limit generalizability and may increase Type II errors. Protocol deviations, unmonitored lab noise, and non-representative noise composition may also impact results.
[208]2019FieldFrom May to November 2016, 105 volunteers from 94 households (mean age 52.1 years) wore wrist actimeters for seven consecutive days to monitor sleep. Environmental and room acoustics were assessed through outdoor and sound measurements.Average noise exposure was not a significant predictor of sleep quality. However, noise during the first 4 h of bedtime reduced sleep latency by 5.6 min per 10 dB(A) increment. Conversely, noise exposure during the final 3 h before waking decreased SE by 2% to 3% per 10 dB(A) increment.Indoor noise was estimated from outdoor measurements, potentially misrepresenting conditions. Censoring noise levels below 20 dB(A) and missing nightly data on window positions contributed to exposure misclassification, reducing statistical power.
[220]2020Lab 50 volunteers (mean age 51.2 years), including 24 living near a wind turbine, participated in a 3-night study: habituation, wind turbine noise exposure, and quiet control. Sleep was assessed with PSG, CORT levels were measured from saliva samples, and subjective experiences were evaluated with morning questionnaires.During exposure nights, REM duration increased by 16.8 min, while the amount of REM sleep decreased by 11.1 min (−2.2%). Other objective sleep measures did not differ significantly between nights. Self-reported sleep quality was consistently worse on exposure nights, and individuals living near wind turbines reported poorer overall sleep compared to the reference group.The study faced potential self-selection bias, as the exposed group included individuals who had complained about wind turbines. The small sample size, deviations from requirements, and laboratory setting may not reflect real-life conditions, introducing bias. Awareness of experimental conditions could also influence self-reported outcomes.
[211]2021SurveyA total of 495 participants completed a questionnaire covering sleep performance, occupational strain, and demographic and lifestyle factors. Nocturnal noise exposure was modeled using road traffic data based on the Nordic prediction method.Nocturnal traffic noise was negatively associated with self-reported poor sleep, while poor sleep was more common in medium- and high-traffic areas. A quiet bedroom façade reduced the incidence of poor sleep, especially at 45–50 dB(A). No difference in poor sleep prevalence was found between windows facing no street and those facing a low-traffic street.The study relied on self-reported sleep data, which may be biased by misclassification and recall issues. It did not directly link disturbances to road noise, affecting accuracy. A restrictive sleep cut-off and lack of OSA screening could have influenced results. Incomplete data on bedroom orientation and noise exposure, with traffic noise as a proxy, may have biased estimates, and missing details on building specifics and other noise sources could have affected findings.
[197]2022Field40 participants, mean age of 29.1 years, were monitored with PSG over five consecutive nights. Indoor sound levels were recorded using a type-1 sound level meter, while road traffic counts were obtained via a radar detector and analyzed into sound levels using custom-developed acoustic software.Traffic noise events were positively associated with REM sleep (P < 0.001) and increased probability of awakening due to maximum noise pressure (P = 0.03) and age (P < 0.005). Conversely, a negative association was observed with SWS (P = 0.017) and the interaction between REM and age (P = 0.017).The small sample size limits generalizability. Study locations with high noise levels may not represent other urban areas. Noise exposure was assessed from traffic counts rather than direct measurements, and the study did not consider confounders such as occupation, income, or temperature.
[204]2022Lab148 patients in intensive care units, scheduled for surgery, with a mean age of 63 years, completed the RCSQ on their first night of admission. Noise levels were monitored over six weeks using a type-2 sound meter.The study found no overall correlation between nighttime sound levels and sleep quality. However, patients with a Visual Analog Scale (VAS) score > 3 or those on hypnotic drugs had lower RCSQ scores (P < 0.05 and P < 0.01, respectively). Excluding these patients showed that each 1 dB(A) increase in nighttime noise was associated with a 3.92-point drop in RCSQ scores (P < 0.05).The study did not account for pre-existing sleep disorders, age, obesity, or disease severity. The sound meter mostly captured staff noise rather than patient-specific sources. Assessments were based on the first night of admission, missing long-term effects, daytime sleep, and circadian disruptions, and lacked patient feedback on perceived sleep disruption causes.
  • The combined effect of ALAN and noise

[83]1999FieldIn a 4-year study of 184 residents (mean age 83.9 years) from eight nursing homes, participants were divided into an intervention group and a control group. The intervention aimed to reduce noise and ALAN exposure. Light and noise levels were recorded with a light meter and microphone, while sleep metrics were collected via actigraphy.While their behavioral intervention resulted in significant reductions in ALAN and noise exposure compared to the control group (P < 0.01), no significant changes were found in sleep patterns.Noise intensity remained high post-intervention, with peaks over 50 dB(A), possibly impacting sleep and reducing effectiveness. The study also overlooked daytime napping, personal factors like medical conditions, and recorded ambient rather than direct exposure, potentially introducing bias.
[223]2016Eco19,136 individuals aged 18 to 99 years completed questionnaires, including self-assessed noise levels. ALAN data were obtained from the Defense Meteorological Satellite Program using NASA’s VIIRS instruments.Increased exposure to outdoor ALAN correlated with delayed bedtimes, later wake-up times, shorter sleep durations, and increased daytime sleepiness (P < 0.01). Intriguingly, noise levels showed no statistically significant correlation with sleep metrics.Self-reported data may suffer from recall and misclassification biases. Relying on self-reported noise and average ALAN measurements could introduce exposure bias. The study also did not analyze the combined impact of ALAN and noise.
[70]2021Eco7,230 individuals with a mean age of 34.5 years completed a national survey focusing on health and well-being, including sleep patterns. Outdoor ALAN exposure was assessed using NASA’s VIIRS data, while noise exposure was interpolated from road density maps.Increased road density and ALAN intensity were associated with reductions in sleep duration (4.5% and 3%, respectively) and increases in sleep difficulty (3.5% and 11%, respectively) (P < 0.01). An interaction effect showed that higher ALAN in noisy areas led to longer sleep but greater sleep difficulty (P < 0.01), possibly due to ALAN penetrating sleeping areas where residents close shades due to heat and humidity.This ecological study combines individual data with environmental estimates, risking misspecification and uncontrolled confounding, which prevents the establishment of causality. Additionally, relying on outdoor ALAN and noise estimates rather than direct indoor measurements may introduce exposure bias.
[229]2021Lab103 patients in the intensive care unit, with a mean age of 53.6 years, completed the RCSQ. Their exposure to noise and ALAN was monitored hourly over one night using a Type 1 noise meter and a light meter.The study revealed a mean noise level of 63.9 dB(A), with peak noise reaching 102 dB(A). The measured ALAN intensity averaged 104.1 lux, peaking at 271 lux. Both stressors were identified as significant predictors of poor sleep (P < 0.001).As a cross-sectional study, causality is unclear. Self-reported data and ambient exposure measurements may introduce bias. Medication, psychological conditions, and the hospital setting might also affect sleep, limiting real-life applicability.
[153]2022EcoData from 51,562 individuals (mean age 66.8 years) in the California Teacher Study cohort were analyzed to evaluate sleep performance. ALAN exposure was assessed using the New World Atlas , and noise levels were derived from the US National Park Service Sound Map. Environmental factors, such as green spaces and air pollution, along with demographic and socioeconomic attributes, were also considered.ALAN and air pollution were linked to shorter sleep durations (< 7 h), while noise was associated with longer sleep onset latency (> 15 min) per 10 dB(A) (P < 0.05). Conversely, green space was associated with increased sleep duration and decreased latency.Self-reported data may be affected by recall and misclassification biases. The study lacks information on the timing of outdoor light exposure and did not consider pre-bedtime device use. Additionally, the US Parks noise model may be less accurate for smaller roadways, and data on ALAN and noise exposure during the night is missing, which could contribute to exposure bias.
[228]2022Survey552 volunteers, with a mean age of 36.7 years, completed questionnaires detailing demographic attributes and reported exposure to ALAN and noise. They also filled out the PSQI, Munich Chronotype Questionnaire, Cognitive Failures Questionnaire, and General Health Questionnaire.The perception of ALAN was significantly associated with poorer sleep quality (P < 0.002), while individuals reporting noise in their bedroom had higher PSQI scores (P = 0.018), suggesting poorer sleep. However, noise within the bedroom showed no significant correlation with sleep duration.As a cross-sectional study, it cannot determine causality. Additionally, self-reported data may introduce recall bias and misclassification. Perceptions of ALAN and noise were not measured directly, which may lead to exposure bias. Moreover, ALAN measurements did not include pre-bedtime exposure.
[71]2023Field72 volunteers, with a mean age of 44 years, from 44 localities in Israel, wore smartwatches to record their sleep metrics and used a smartphone to log exposure to ALAN and noise for 30 consecutive nights. Diaries recorded subjective data, and questionnaires were used to gather demographic, behavioral, habitual, and health information.Exposure to ALAN and noise before bedtime significantly reduced sleep duration by 9.7% to 9.9% and SE by 2.4% to 2.8% (P < 0.01). While ALAN exposure during sleep had no significant impact, noise was notably disruptive, reducing sleep duration by 13.5% and SE by 4.4%. The combined effect of ALAN and noise led to a substantial decrease in sleep metrics, with reductions of 15.3% in duration and 8% in SE before sleep, and 14% and 9%, respectively, during sleep. The results suggest that ALAN and noise may amplify each other’s effects, potentially worsening health outcomes.The study’s limitations include its short duration (one month) and single-country scope, which may limit generalizability. Most participants were under 60, potentially overlooking age-related impairments. Variability in personal smartphones and smartwatches, as well as the lack of seasonal and spectral considerations for light and noise, may also affect results.

ALAN: artificial light at night; CORT: cortisol; Eco: ecological; EEG: electroencephalographic; WASO: wakes after sleep onset; HR: heart rate; ipRGCs: intrinsic photosensitive retinal ganglion cells; S: shor wavelengths of light; L: long wavelengths of light; M: medium wavelengths of light; Lab: laboratory; MLT: melatonin; NREM: non-rapid-eye-movements; N1–N3: NREM 1 to 3 sleep stages; OR: odd ratio; OSA: obstructive sleep apnea; PSG: polysomnography; SWL: short wavelength light; PSQI: Pittsburgh sleep quality index; RCSQ: Richards-Campbell sleep questionnaire; REM: rapid eye movements; Retro: retrospective; SE: sleep efficiency; LED: light emitted diode; SWS: slow-wave sleep; NA: noradrenaline; VIIRS: Visible Infrared Imaging Radiometer Suit; 1 Arranged from oldest to newest and excludes reviews, letters, or books, focusing primarily on studies investigating the relationship with circadian cycles, sleep, and MLT; 2 Ref numbers are organized in the order they appear in the text; 3 The category “laboratory” covers research conducted in medical institutions, research facilities, and controlled clinical settings, whereas the “ecological” category includes both individual-based and population-based epidemiological studies