Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Sep 1.
Published in final edited form as: Sleep Med. 2024 Jun 14;121:69–76. doi: 10.1016/j.sleep.2024.06.013

Long-term Night-Shift Work is Associated with Accelerates Brain Aging and Worsens N3 Sleep in Female Nurses

Soonhyun Yook 1,†, Su Jung Choi 2,†, Hanul Lee 3, Eun Yeon Joo 4, Hosung Kim 1
PMCID: PMC11330713  NIHMSID: NIHMS2006016  PMID: 38936046

Abstract

Background:

Shift work disrupts circadian rhythms and alters sleep patterns, resulting in various health problems. To quantitatively assess the impact of shift work on brain health, we evaluated the brain age index (BAI) derived from sleep electroencephalography (EEG) results in night-shift workers and compared it with that in daytime workers.

Methods:

We studied 45 female night shift nurses (mean age: 28.2±3.3 years) and 44 female daytime workers (30.5±4.7 years). Sleep EEG data were analyzed to calculate BAI. The BAI of night shift workers who were asleep during the daytime with those of daytime workers who were asleep at night were statistically compared to explore associations between BAI, duration of shift work, and sleep quality.

Results:

Night-shift workers exhibited significantly higher BAI (2.14±6.04 vs. 0±5.35), suggesting accelerated brain aging and altered sleep architecture, including reduced delta and sigma wave frequency activity during non-rapid eye movement sleep than daytime workers. Furthermore, poor deep sleep quality, indicated by a higher percentage of N1, lower percentage of N3, and higher arousal index, was associated with increased BAI among shift workers. Additionally, a longer duration of night-shift work was correlated with increased BAI, particularly in older shift workers.

Conclusion:

Night-shift work, especially over extended periods, may be associated with accelerated brain aging, as indicated by higher BAI and alterations in sleep architecture. Interventions are necessary to mitigate the health impacts of shift work. Further research on the long-term effects and potential strategies for sleep improvement and mitigating brain aging in shift workers is warranted.

1. Introduction

Shift work that encompasses shifts outside the hours of 7 a.m. to 6 p.m. and constitutes fixed or rotating schedules, including morning, evening, or night shifts (Redeker et al., 2019), accounts for 15–25% of workers globally.(Wickwire et al., 2017) Shift work disrupts one’s circadian rhythms and alters normal activities such as eating and sleeping patterns, resulting in various physical and mental health issues (Kecklund & Axelsson, 2016). It increases the risk of developing physical diseases such as cerebrovascular disease, metabolic syndrome, obesity, and cancer, and negatively impacts mental health such as depression, mood disorders, and suicide (Kecklund & Axelsson, 2016; Torquati et al., 2019; Wyse et al., 2017). An epidemiologic study of >27,000 UK workers has revealed that shift workers had more health problems such as sleep disorders, obesity, and depression, despite being more physically active than fixed daytime workers (Wyse et al., 2017).

Among these various health issues, poor-quality sleep resulting from circadian misalignment is known to affect not only shift workers’ alertness on duty (Alfonsi et al., 2021; Hong et al., 2021) but also their cognitive function (Esmaily et al., 2022). Sleep problems are particularly severe in night-shift workers, who struggle to maintain quality daytime sleep because of circadian misalignment (Alfonsi et al., 2021; Wright Jr et al., 2013). Moreover, daytime sleep in this type of work is more fragmented than normal nighttime sleep (Chang & Li, 2019). Sleep fragmentation can reduce sleep spindle and slow wave, which are involved in memory consolidation (Mander et al., 2017). Furthermore, the resulting reduction in sleep duration provides insufficient time for complete memory consolidation (Joiner, 2019).

The long-term consequences of shift work, particularly night shifts, have become increasingly evident. Prolonged exposure to night-shift work is associated with a higher risk of various diseases, including cancer, heart disease, and stroke. Additionally, the interplay between increased sleep fragmentation and circadian rhythm disruption may contribute to neurodegenerative diseases such as Alzheimer’s disease (Musiek et al., 2015) and cognitive impairment (Lee et al., 2023).

Recent studies have developed a quantitative manner to measure brain health using sleep electroencephalography (EEG), called the brain age index (BAI). This method has been successful in predicting the risk of neurodegenerative diseases, psychiatric disorders, cognitive decline, and even mortality (Paixao et al., 2020; Sun et al., 2019; Ye et al., 2020). More recently, our group advanced this method by employing a deep learning approach, improving the ability to more accurately estimate brain age. We found that poor sleep quality, stemming from conditions such as sleep apnea and insomnia, can accelerate brain aging (Yook et al., 2022). However, up to date, the specific impact of long-term circadian rhythm alterations, as experienced in shift work, on brain aging remains underexplored.

This study aimed to assess the relationship between sleep EEG-based BAI and shift work. By comparing daytime workers with nighttime sleep and shift workers with daytime sleep, we sought to identify factors that might be associated with accelerated brain aging in night-shift workers. We also investigated the role of sleep EEG patterns in potentially hastening the BAI among night-shift workers.

2. Methods

2.1. Participants and study screening

Participants were recruited from two cohorts at the Samsung Medical Center in Seoul, South Korea: (a) night-shift work(SW) nurses with daytime sleep and (b) daytime work (DW) participants with regular nighttime sleep. We advertised in the hospital cafeteria to recruit participants. The exclusion criteria were current use of drugs (i.e., tobacco, alcohol >2 glasses per day, benzodiazepines), use of potential sleep-altering medications, and history of treatment for sleep, medical, and psychiatric disorders. All participants who voluntarily enrolled in this study provided written informed consent. All procedures were approved by the Institutional Review Board of Samsung Medical Center (IRB no: 2018-05-120 for SW nurses with daytime sleep and 2018-10-037 for DW participants with nighttime sleep) and conducted in accordance with the Declaration of Helsinki.

2.1.1. Night-shift work nurses with daytime sleep

Forty-eight healthy female nurses working in shifts were enrolled in this study. Eligible participants had rotating shift work careers, including night shifts for at least 1 year. All participants were working 12-hour shifts, and had the same shift pattern of Day-Day-Night-Night shift, with 4 consecutive days on duty followed by 4-5 days off. The working hours of the day shift were from 7:00 a.m. to 7:30 p.m., and those of night shift were from 7:00 p.m. to 7:30 a.m., including a 30-minute mealtime in each shift. Participants underwent in-lab polysomnography after their 2nd night shift. Three participants were excluded because of unmet work schedule requirements. In total, 45 nurses (mean 28.2±3.3 years) were included in the study.

2.1.2. DWs with regular nighttime sleep

In total, 44 healthy female adults (mean 30.5±4.7 years) were involved with the intermediate chronotype as defined by a morningness–eveningness questionnaire (Horne & Ostberg, 1976) and 7-day sleep diary. The participants rested between 11:00 p.m. and 8:00 a.m. The cohort was recruited and analyzed as previously described.(Jo et al., 2021; H. R. Park et al., 2020)

2.2. Sleep EEG data acquisition

Sleep polysomnography (PSG) was conducted using the Embla N7000 system (Embla, Reykjavik, Iceland). Data were collected from eight EEG channels, including frontal channels (F3, F4), central channels (C3, C4), occipital channels (O1, O2), and channels located behind the ears (A1, A2), accompanied by an electrooculogram, electromyogram, and electrocardiogram. All recordings were conducted at a sampling rate of 200 Hz.

For the SW group, the participants typically went to bed between 9:00 a.m. and 10:00 a.m., with wake-up times ranging from 12:00 p.m. to 5:30 p.m. The beginning of the sleep study and lights off time were aligned with the regular sleeping hours of participants following a 12-hour night shift. The PSG study was considered complete when participants woke up naturally. If a participant did not awaken by 5:00 p.m., they were awakened by a technician at 5:30 p.m., marking the end of the PSG study. Participants then reported their self-estimated sleep (subjective total sleep time (TST) and sleep latency). In the DW group, participants entered the sleep laboratory around 4:00 p.m., had a meal, and then applied the PSG electrodes. They stayed awake in a chair under exposure to 3,000 or 4,000 K LED light from 6:00 p.m. until midnight. Subsequently, participants went to bed according to their usual setup and woke up the following morning at their regular time for work.

2.3. EEG preprocessing and artifact removal

To prepare the data, we applied a bandpass filter (0–50 Hz) and eliminated the artifacts associated with the electrocardiogram and electrooculogram signal components using the automatic artifact removal plugin of the EEG LAB. Subsequently, we assessed the signal amplitude for each EEG channel at every time point for all the participants. Any amplitude value exceeding five standard deviations was identified as an artifact and corrected by interpolation using neighboring signals based on spline interpolation. This interpolation process is required for approximately 3.8% of the data.

Channels with artifact proportions of >30% across the EEG data were categorized as “bad” channels, although such instances were not present in our dataset. Finally, z-score normalization was performed by standardizing the amplitude of each EEG channel relative to the mean and standard deviation of the entire dataset.

2.4. EEG relative power analysis

In our study, we utilized MNELAB, a Python-based processing toolbox, to analyze the spectral patterns of the six-channel sleep EEG data (Gramfort et al., 2014). We divided the data into five frequency bands and computed the power for each band as follows: delta (0.5–4 Hz), theta (4–8 Hz), alpha (8–12 Hz), sigma (12–15 Hz), and beta (15–30 Hz). The sleep stages (N1–N3, REM) were manually scored according to the guidelines outlined by the American Academy of Sleep Medicine standards (Berry et al., 2012). Subsequently, we calculated the spectral pattern of the sleep EEG data separately for each sleep stage. To enhance the spectral characteristics, we averaged the power values of the six-channel EEGs to obtain a single value. Finally, we determined the relative EEG power for each sleep stage by calculating the ratio of power within a specific frequency band to the total power (0.5–15 Hz) of each participant.

2.5. Calculating BAI

To predict BAI (details in S1 and Figure S1), we utilized a previously developed deep learning model that can predict brain age from sleep EEG recording(Yook et al., 2022). We trained this computational model on data from healthy subjects with normal sleep to predict how old their ages were at sleep EEG recording (= chronological age). During model training, we hypothesized that healthy sleeper’s brain age equates with their chronological age. After training, the model could measure electrophysiological brain aging in unseen subjects. For example, this model was tested in patients with obstructive sleep apnea and revealed faster brain aging in this patient group (i.e. predicted age was older than the subject’s chronological age).

The rationale that the model can extract aging information from sleep EEG is based on the fact that the structure of sleep stages and cycles alters along with aging. Also, the power spectrum of the different frequency waves seen during sleep alters along with aging. For example, with aging, the power of delta and theta waves and the number of sleep spindles tend to decrease whereas alpha power increases. We thus input the following two EEG features into the model: 1) sleep stage resampled to 2000-time bins, forming a 2000-bin by 1 × 1 matrix and; 2) scalogram that represents power spectrum changes over time. This data was generated by transforming one-dimensional (1D) sleep EEG data into two-dimensional (2D) images using a continuous wavelet transform technique, which converted the 1D EEG signals from each of six channels into 2D images including 2000 x16 pixels. In these scalogram images, the y-axis represents frequency (bottom to top mean slowest to fastest frequency), the x-axis represents time (each pixel represents approximately 10 seconds), and each pixel’s color intensity means the power level of the signal (e.g. a high intensity value implies a larger amplitude and/or a more number of the given frequency wave seen at the time point). During the training, the model learns the pattern of each individual’s sleep stage and various frequency wave power alterations over time and fit it into the subject’s age via nonlinear regression. Therefore, the model turned the input sleep stage and scalogram data into an age value. We used a deep learning neural network algorithm, namely DenseNet, as the prediction model.

We then calculated the BAI per participant, which is an indicator of the participant’s relative brain health status, by subtracting their chronological ages from their predicted brain ages (Cole & Franke, 2017; Jonsson et al., 2019; Ning et al., 2021; Smith et al., 2019). More technical details are found in our previous study (Yook et al., 2022).

2.6. Statistical analysis

2.6.1. Comparison between daytime sleep of a SW and nighttime sleep of a DW groups

To determine whether shift work is associated with brain aging, we compared the BAI between SW with daytime sleep and DW with nighttime sleep groups while controlling for relevant covariates, ultimately providing insights into the impact of shift work. A linear regression analysis was performed. To ensure a robust comparison and account for potential confounding factors, we incorporated several covariates including chronological age and body mass index (BMI).

2.6.2. Regional spectral power analysis

To evaluate whether the potential increase in BAI in the SW group could be explained by the different patterns of the regional EEG spectra, we analyzed the spatial characteristics of the EEG spectral power for NREM across six channels and five frequency bands. The spectral powers measured for the two channels in the frontal, central, and occipital lobes were averaged to interpret the findings with respect to lobes. Subsequently, we compared the spatial characteristics of the spectral power for each sleep disorder group with those of healthy sleepers by performing linear regression independently at each frequency band and each lobe while adjusting for chronological age and BMI.

Furthermore, we split our samples using a median chronological age of 29.3 years: those aged ≥29.3 years (N = SW: 9, DW: 26) vs. those <29.3 years (N = SW: 36, DW: 19). Within these age-based subgroups, we compared daytime sleep in the SW group with nighttime sleep in the DW group using the aforementioned method.

2.6.3. Sleep cycle separation

We analyzed the changes in EEG spectral patterns for N3 sleep and the sleep structure corresponding to the progression of sleep between SW and DW groups. Therefore, we segmented the sleep cycle, which initiates with NREM sleep and ends with transitions into REM sleep as the sleep becomes more profound (Carskadon & Rechtschaffen, 2011). We developed a Matlab toolbox for the sleep cycle partition (GitHub link: https://github.com/ZCONG2025/Sleep-Cycle-Seperation/tree/main) to visualize the sleep stages of the participants and perform a single click on the hypnogram to record split points of different cycles.

In the SW group, daytime sleep demonstrated an average of 4.05±0.82 sleep cycles, ranging from 3 to 6 cycles. By contrast, the DW group exhibited an average of 4.62±0.88 sleep cycles during nighttime sleep, ranging 3 to 6 cycles. As the dataset had a minimum of three sleep cycles, and considering variations in the number of cycles among individuals, we extracted the first, second, and final sleep cycles from all the participants. The selected cycles were then used to analyze the EEG temporal patterns. Finally, we compared the N3 sleep (%) and EEG delta relative power of N3 sleep for each of the three segmented sleep cycles between the SW and DW groups using linear regression while adjusting for chronological age and BMI.

2.6.4. Association of various sleep parameters with BAI

To quantitatively assess the effects of sleep parameters and other covariates on increased BAI, we used a linear regression model. Regression analyses were performed separately between BAI and each parameter/variable for the SW and DW groups. The sleep and non-sleep variables analyzed using the linear model included the following: age, BMI, Epworth sleepiness scale (ESS), insomnia severity index (ISI), shift work starting age, shift working period (month), arousal index (AI), apnea hypopnea index (AHI), oxygen desaturation index, NREM (%), N1(%), N2 (%), N3 (%), and REM (%).

2.6.5. Associations between shift work duration and BAI and between age and BAI

To investigate whether the BAI during daytime sleep in the SW group increased with age compared to that during nighttime sleep in the DW group, we conducted a linear regression analysis to examine the interaction between group and age.

Furthermore, we assessed the possible linear or nonlinear relationships between shift work duration and BAI within the SW group. We also assessed how EEG spectral patterns are associated with the increase in the BAI as the duration increased. To explore this relationship, we employed linear regression to examine the interaction between shift work duration and the relative power of delta, theta, alpha, sigma, and beta frequencies during NREM and REM sleep in relation to the BAI.

To maintain statistical validity, all statistical results were adjusted for multiple comparisons by controlling for the false discovery rate.

3. Results

3.1. Demographics and clinical characteristics

BMI, ESS scores, AI, wake after sleep onset (WASO), sleep efficiency, and proportions of NREM, N1, and REM sleep did not exhibit significant differences between the nighttime sleep DW group and the daytime sleep SW group. However, the SW group experienced a shorter TST (299.8±81.7 min) compared to the DW group (374.0±61.8 min, p < 0.0001), a shorter sleep latency (2.5±3.2 min vs. 11.5±23.7, p = 0.01), a shorter REM latency (64.0±33.1 min vs. 91.3±48.1, p = 0.002), a lower percentage of N2 sleep (44.6±9.3% vs. 50.8±8.6%, p = 0.002), and a higher percentage of N3 sleep (23.8±8.1% vs. 15.0±10.5%, p = 0.0001). Additionally, the SW group was younger (28.2±3.3 years vs. 30.5±4.7, p = 0.008) and had a lower BMI (20.1±1.8 vs. 22.0±4.1, p = 0.006) than the DW group (Table 1).

Table 1.

Demographics and clinical characteristics

Nighttime sleep DW
(n=44)		 Daytime sleep SW
(n=45)		 p-value
Age, yr 30.5 (4.7) 28.2 (3.3) 0.008*
BMI, kg/m2 22.0 (4.1) 20.1 (1.8) 0.006*
ESS 9.3 (4.7) 9.7 (3.8) 0.70
AHI 1.6 (2.2) 0.6 (1.4) 0.06
AI, /h 15.1 (7.2) 12.6 (5.6) 0.07
TST, min 374.0 (61.8) 299.8 (81.7) <0.0001*
Sleep latency, min 11.5 (23.7) 2.5 (3.2) 0.01*
REM latency, min 91.3 (48.1) 64.0 (33.1) 0.002*
WASO, min 30.3 (22.0) 34.4 (23.7) 0.41
Sleep efficiency, % 90.2 (7.9) 89.0 (6.6) 0.44
NREM, % 77.7 (5.0) 78.5 (5.9) 0.49
N1, % 11.9 (6.7) 10.9 (5.0) 0.41
N2, % 50.8(8.6) 44.6 (9.3) 0.002*
N3, % 15.0(10.5) 23.8 (8.1) 0.0001*
REM, % 22.3 (5.0) 21.5 (5.9) 0.49
Open in a new tab

Abbreviations: BMI, Body mass index; ESS, Epworth sleepiness scale; AHI, Apnea hypopnea index; AI, Arousal index; TST, Total sleep time; WASO, Wake after sleep onset; NREM, Non-rapid eye movement sleep; REM, Rapid eye movement sleep.

3.2. High BAI in daytime sleep of SWs

To investigate whether SW group is associated with accelerated BAI, we compared the mean BAI of the SW and DW groups. The mean BAI for the SW group (mean±SD: 2.14±6.04 years) was 2.14 years higher than that of the sleep DW group (0±5.35 years), as shown in Figure 1A, with p < 0.05 after adjusting for age and BMI. Hence, the analysis of sleep EEG spectral patterns (Figure 1B) revealed that the SW group exhibited lower power in the range of 0.5–15 Hz, encompassing delta, theta, alpha, and sigma waves in the frontal lobe, than the DW group.

Figure 1. Comparison of BAI and EEG power between DW and SW groups.

Figure 1.

Open in a new tab

A. The SW group exhibits a significantly higher BAI compared to the DW group, The box and whisker plot displays the upper quartile (light box), the lower quartile (the dark box), and the median (the line between these two boxes). B. The SW group demonstrates significantly lower EEG power in the delta, theta, alpha, and sigma frequencies in the frontal lobe than the DW group.

DW, day worker; SW, shift worker; BAI, brain aging index; EEG, electroencephalogram.

3.3. Comparison of temporal patterns of deep sleep-related parameters between the DW and SW groups

To investigate whether SW is associated with sleep architecture, especially deep sleep, we analyzed the temporal patterns of deep sleep-related parameters, such as the percentage of N3 sleep and relative delta wave power during N3, which may vary with the sleep cycle in different groups. The percentage of N3 sleep in the SW group was significantly higher than that in the DW group in the first sleep cycle (p = 0.01) but not in the other cycles (Figure 2A). By contrast, the relative delta wave power during N3 sleep in SW group was significantly lower in the last sleep cycle than that in the DW group (p = 0.03, Figure 2B), whereas no significant difference was observed between the two groups during the first and second sleep cycles. On the other hand, the absolute delta power in N3 was lower in the SW group compared to the DW group across all sleep cycles (Figure S2).

Figure 2. Comparison of temporal patterns of N3 sleep (%) and N3 delta power between the SW and DW groups.

Figure 2.

Open in a new tab

A. In the first sleep cycle, the SW group exhibits higher N3 sleep (%) than the DW group. B. In the last sleep cycle, the DW group exhibits higher relative power of delta wave during N3 sleep than the SW group. DW, day worker; SW, shift worker.

3.4. Association between participant’s characteristics and BAI

The coefficients of the covariates in our linear regression model, which represent the estimated relationship between each variable and BAI, are presented in Figure 3. In the SW group, the following covariates were significantly associated with a higher BAI; chronological age had the highest positive regression coefficient with BAI (regression coefficient ± standard error: 0.9 ± 0.3; p < .001), followed by AI (0.4 ± 0.2; p = .013), percentage of N1 sleep (0.4 ± 0.2; p =.007), and duration of shift work (0.1 ± 0.03; p < .001). The percentage of N3 sleep (−0.3 ± 0.1; p =.003) was negatively correlated with BAI. In the DW group, no significant association was observed between any of the variables and BAI.

Figure 3. Regression analysis of BAI.

Figure 3.

Open in a new tab

In the SW group, chronological age, AI, N1 sleep (%), and duration of shift work (month) demonstrates a positive correlation with BAI. Conversely, N3 sleep (%) displays a negative correlation with BAI. In the DW group, no variable exhibited a significant association with BAI. DW, day worker; SW, shift worker; BAI, brain aging index; AI, arousal index, AHI, apnea hypopnea index; ODI, oxygen desaturation index; ISI, insomnia severity index; BMI, body mass index; ESS, Epworth sleepiness scale.

3.5. Association between BAI and chronological age in the SW group

We examined whether a significant quadratic association exists between an increase in the BAI and chronological age in the daytime sleep SW group (red line in Figure 4-bottom, p < .001). Furthermore, a significant interaction in BAI change was observed between chronological age and groups (SW vs. DW) (Figure 4-top, p = 0.005). Specifically, the SW group demonstrated a higher BAI in older participants, whereas no similar trend was observed in the DW group. In line with this, among those aged ≥30 years, the SW group exhibited a significant reduction in spectral power across delta, theta, and sigma waves during NREM sleep compared to the DW group (Figure 4-bottom). Conversely, for those aged <30 years, no significant spectral power differences were observed between the two groups.

Figure 4. An elevated BAI was observed with aging participants in the SW group compared to the DW group.

Figure 4.

Open in a new tab

Among participants aged ≥29.3 years, the SW group displays significantly reduced spectral power in delta, theta, and sigma waves during non-rapid eye movement sleep compared to the DW group. However, no significant differences are observed between groups for participants aged <29.3 years. BAI, brain aging index; SW, shift worker; DW, day worker.

3.6. Association between BAI and shift work duration

Our evaluation of the relationship between shift work duration and BAI revealed a longer shift work duration was associated with a higher BAI (r = 0.36, p = 0.001) (Figure 5). Our interaction analysis also assessed which EEG components contributed to the increase in the BAI alongside an increase in shift work duration. Only the delta wave activity during NREM sleep made a significant contribution.

Figure 5. Association between duration of shift work and BAI.

Figure 5.

Open in a new tab

A longer duration of shift work has a quadratic relationship with a higher BAI. Additionally, the interaction between shift work duration and delta power in non-stage R sleep is significant in relation to changes in BAI. BAI, brain aging index.

4. Discussion

Our study analyzed the impact of shift work on brain health using the EEG-based BAE The BAI serves as an index of brain health that can be reflected in each sleep session. However, the current study has a limitation in its interpretation because the BAI was not equally from overnight polysomnography results for both night SW and DW group. Thus, it is unclear whether the difference in BAI between these two groups is due to immediate effects of one daytime sleep or chronic effects of a long-term SW. Given the reduced delta and sigma wave frequency activity during NREM sleep found in the night SW group, the increase in BAI for night shift workers compared to day workers could imply alterations in their sleep architecture during daytime sleep. Nevertheless, among night shift workers whose BAI was measured from daytime sleep, we discovered that individuals with a longer duration of shift work exhibited a greater increase in BAI, which may be related to the chronic effects of long-term circadian rhythm disruption and the subsequent deprivation of deep sleep.

4.1. Night-shift work and daytime sleep accelerate brain aging

The increased brain age observed in night-shift workers compared to DWs suggests an acceleration of neuroelectrophysiological aging processes accompanied with decreased delta and sigma wave activity, which are essential for cognitive restoration and memory consolidation during NREM sleep (BuzsÁk, 1998; Holz et al., 2012). This adverse process may be explained by circadian rhythm shift as well as commonly observed poor deep sleep owing to sleep misalignment (Hasan et al., 2018) and sleep fragmentation (Ramesh & Gozal, 2009) of daytime sleep.

Previous studies have reported that shift work negatively affects the brain health. After night shifts, nurses demonstrated reduced activity in the prefrontal cortex and a decline in cognitive performance (Duran-Gomez et al., 2021). Another study has highlighted a decrease in executive function (Athar et al., 2020). A mouse model confirms that circadian misalignment can disrupt glymphatic function, potentially leading to the accumulation of toxic metabolites in brain tissues and cognitive decline (Hablitz et al., 2020).

In this study, an intriguing finding is the higher percentage of N3 sleep (>8.8%) and shorter sleep latency (<9min) observed in SW group during daytime sleep compared to DW group during nighttime sleep. This pattern is more evident during the initial sleep cycle than during other cycles, suggesting the increased homeostatic drive for sleep that results from sleep deprivation during night shift work, which may further lead to decreases in sleep latency and increases in the percentage of N3 sleep, especially in the early cycles of sleep (Diane B Boivin et al., 2022). On the other hand, relative delta power was notably reduced during N3 sleep in the latter parts of daytime sleep for SW group. Conversely, other frequency bands (theta, alpha, sigma, and beta) showed opposite trends to delta, but the differences were not statistically significant. This may be due to the depth or quality of this deep sleep phase being less optimal in daytime sleep after SW (Yook et al., 2024).

Exposure to light during the night shift leads to melatonin suppression (D. B. Boivin et al., 2022; Rahman et al., 2017), and melatonin suppression can cause reduced delta power during NREM in the following sleep (Rahman et al., 2017). Indeed, melatonin suppression in night shift workers has been reported to be lower than that of day workers, not only in daytime sleep following a night-shift, but also in nighttime sleep on off-nights (Mirick et al., 2013). Therefore, decrease in NREM delta power led by melatonin suppression in night shift workers may be present during both their daytime sleep following shifts and nighttime sleep on off-nights.

Our study found that the high BAIs in shift workers were linked to delta power reduction. The reduction in delta power in shift workers are likely attributed to the misalignment of circadian rhythms and chronic sleep deprivation, both of which have been associated with neurodegeneration (Abbott & Videnovic, 2016; Leng et al., 2019; Özdemir et al., 2013). This may serve as evidence that reduced deep sleep quality can lead to accelerated brain aging in shift workers.

As depicted in Figure 3, the increase in BAI in the SW group indicated a correlation with an increase in the AI and N1 sleep (%) as well as a decrease in N3 sleep (%), implying that the poor quality of sleep, in terms of shallower sleep and more frequent awakenings during sleep, in SW group is possibly linked to accelerated brain aging. In previous studies, poor quality sleep of shift workers has been demonstrated to have associations with brain health. For example, sleep disturbances in nurses working shifts have been associated with reduced brain volume (Bittner et al., 2022; Kim & Kim, 2017; C. Park et al., 2020) and a decline in cognitive function (Athar et al., 2020; Kazemi et al., 2016; Özdemir et al., 2013). Our results also align with these findings when assessing brain health outcomes using the EEG-BAI, demonstrating the ability of quantitative brain aging metrics to assess the impact of shift work on brain health.

4.2. Longer duration of shift work is associated with accelerated brain aging

In this study, older age correlated with increased BAI, whereas the age at which an individual started shift work did not indicate a correlation with BAI. However, a longer duration of shift work was associated with a higher BAI. Furthermore, the combination of long-term shift work and a decrease in delta waves during NREM sleep synergistically contributed to further increases in BAI. Rouch et al. have reported that memory performance decreases with increasing shiftwork duration (Rouch et al., 2005). Another previous study has indicated that compared with people who had <5 years of shiftwork exposure, immediate recall performance decreased as shiftwork experience increased, and significantly so between 10 and 20 years, which is similar to our findings. In addition, Marquie et al. (Marquie et al., 2015) have reported in the VISAT (aging, health, and work) prospective study that exposure to shift work was associated with chronic impairment of cognition, and the association was highly significant for exposure to rotating shiftwork exceeding 10 years. Altogether, the previous findings and our results suggest that an increase in BAI due to long-term circadian misalignment may be cumulative.

4.3. Limitations

The current study has several limitations. First, the sample size was relatively small, with a total of 89 participants, including both SW and DW groups. Additionally, the data used in this study were from a single site, the Samsung Medical Center. To generalize the findings, further research involving a larger sample size and data from multiple sites is necessary. Second, we did not include fixed daytime nurses. As senior nurses are typically assigned fixed daytime shifts in South Korea (Jeong et al., 2019), this could have biased the age distribution in the study samples. Brain age is significantly influenced by biological age, and this effect may be stronger among nurses. In addition, fixed daytime nurses have some experience with rotating shift work; therefore, the effect of shift work experience cannot be completely ruled out. Therefore, we included non-nursing female daytime workers as controls. Third, we recruited only female participants because of the disproportionate female population in the nursing profession. The generalizability of this study may be improved by including male participants, other age groups, fixed-night-shift workers, and other occupations. Fourth, the analysis of the effects of shiftwork duration was based only on cross-sectional sample data, limiting our understanding of the possibly individually varying courses of shiftwork experience. Fifth, a larger sample is required to assess whether resilience to circadian rhythm misalignment varies among different chronotypes. Sixth, this study was unable to disentangle the effects of the workload of night-shift work, sleep deprivation during the night, circadian rhythm shift, exposure to light during nighttime work, and poor sleep quality during daytime sleep, although these factors as a whole can contribute to the observed changes in EEG power. Finally, this study observed an increase in BAI during daytime sleep following night shifts. However, since BAI was originally designed to analyze brain aging during nighttime sleep, it is limited in directly proving evidence of brain aging from these changes. To better assess the impact of repeated poor-quality sleep due to night shift work on brain health, future studies should analyze BAI during nighttime sleep in shift workers.

4.4. Conclusions

This study highlights the potential association between night-shift work and brain aging. Our findings emphasize the importance of occupational brain health interventions that address the unique challenges posed by shiftwork. In particular, the disruption of deep sleep during daytime sleep and prolonged shift work can synergistically accelerate brain aging. Future research should focus on broader demographics and adopt longitudinal designs to elucidate the long-term effects associated with shift work and the potential reversibility of brain aging after returning to regular daytime work. Developing effective strategies to mitigate the adverse effects of shift work on brain health is crucial, particularly for individuals engaged in long-term shift work.

Supplementary Material

1

Highlight.

  • Night-shift workers exhibit a significantly higher Brain Age Index (BAI) compared to daytime workers.

  • Deteriorated deep sleep quality during daytime sleep for night-shift workers, characterized by increased N1 stages, reduced N3 stages, and elevated arousal index, is linked to an increased BAI.

  • Extended periods of night-shift work are associated with accelerated brain aging.

Acknowledgments

This research was supported by grants from the National Institutes of Health (P41EB015922) and Samsung Medical Center (OTC1190671).

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Alfonsi V, Scarpelli S, Gorgoni M, Pazzaglia M, Giannini AM, & De Gennaro L (2021). Sleep-Related problems in night shift nurses: towards an individualized interventional practice. Frontiers in Human Neuroscience, 15, 644570. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Archer SN, Laing EE, Moller-Levet CS, van der Veen DR, Bucca G, Lazar AS, … Dijk DJ (2014). Mistimed sleep disrupts circadian regulation of the human transcriptome. Proc Natl Acad Sci U S A, 111(6), E682–691. 10.1073/pnas.1316335111 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Athar ME, Atef-Vahid M-K, & Ashouri A (2020). The influence of shift work on the quality of sleep and executive functions. Journal of circadian rhythms, 18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Berry RB, Brooks R, Gamaldo CE, Harding SM, Marcus C, & Vaughn BV (2012). The AASM manual for the scoring of sleep and associated events. Rules, Terminology and Technical Specifications, Darien, Illinois, American Academy of Sleep Medicine, 176, 2012. [Google Scholar]
  5. Bittner N, Korf H-W, Stumme J, Jockwitz C, Moebus S, Schmidt B, … Caspers S (2022). Multimodal investigation of the association between shift work and the brain in a population-based sample of older adults. Scientific Reports, 12(1), 2969. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Boivin DB, Boudreau P, & Kosmadopoulos A (2022). Disturbance of the Circadian System in Shift Work and Its Health Impact. J Biol Rhythms, 37(1), 3–28. 10.1177/07487304211064218 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Boivin DB, Boudreau P, & Kosmadopoulos A (2022). Disturbance of the circadian system in shift work and its health impact. Journal of Biological Rhythms, 37(1), 3–28. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. BuzsÁk G. (1998). Memory consolidation during sleep: a neurophysiological perspective. Journal of sleep research, 7(S1), 17–23. [DOI] [PubMed] [Google Scholar]
  9. Carskadon MA, & Rechtschaffen A (2011). Monitoring and staging human sleep. Principles and practice of sleep medicine, 5, 16–26. [Google Scholar]
  10. Chang W-P, & Li H-B (2019). Differences in workday sleep fragmentation, rest-activity cycle, sleep quality, and activity level among nurses working different shifts. Chronobiol Int, 36(12), 1761–1771. [DOI] [PubMed] [Google Scholar]
  11. Cole JH, & Franke K (2017). Predicting Age Using Neuroimaging: Innovative Brain Ageing Biomarkers. Trends Neurosci, 40(12), 681–690. 10.1016/j.tins.2017.10.001 [DOI] [PubMed] [Google Scholar]
  12. Duran-Gomez N, Guerrero-Martin J, Perez-Civantos D, Lopez-Jurado CF, Montanero-Fernandez J, & Caceres MC (2021). Night Shift and Decreased Brain Activity of ICU Nurses: A Near-Infrared Spectroscopy Study. Int J Environ Res Public Health, 18(22). 10.3390/ijerph182211930 [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Esmaily A, Jambarsang S, Mohammadian F, & Mehrparvar AH (2022). Effect of shift work on working memory, attention and response time in nurses. International Journal of Occupational Safety and Ergonomics, 28(2), 1085–1090. [DOI] [PubMed] [Google Scholar]
  14. Gramfort A, Luessi M, Larson E, Engemann DA, Strohmeier D, Brodbeck C, … Hämäläinen MS (2014). MNE software for processing MEG and EEG data. Neuroimage, 86, 446–460. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Hablitz LM, Plá V, Giannetto M, Vinitsky HS, Stæger FF, Metcalfe T, … Nedergaard M (2020). Circadian control of brain glymphatic and lymphatic fluid flow. Nature Communications, 11(1), 1–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Hasan S, Foster RG, Vyazovskiy VV, & Peirson SN (2018). Effects of circadian misalignment on sleep in mice. Scientific reports, 8(1), 1–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Holz J, Piosczyk H, Feige B, Spiegelhalder K, Baglioni C, Riemann D, & Nissen C (2012). EEG sigma and slow-wave activity during NREM sleep correlate with overnight declarative and procedural memory consolidation. Journal of sleep research, 21(6), 612–619. [DOI] [PubMed] [Google Scholar]
  18. Hong J, Choi SJ, Park SH, Hong H, Booth V, Joo EY, & Kim JK (2021). Personalized sleep-wake patterns aligned with circadian rhythm relieve daytime sleepiness. Iscience, 24(10). [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Horne JA, & Ostberg O (1976). A self-assessment questionnaire to determine morningness-eveningness in human circadian rhythms, Int J Chronobiol, 4(2), 97–110. [PubMed] [Google Scholar]
  20. Jeong JL, Kwon HM, Kim TH, Cho i. M. R., & Eun HJ (2019). Effects of Perceived Stress, Sleep, and Depression on Resilience of Female Nurses in Rotating Shift and Daytime Fixed Work Schedules. Sleep Med Psychophysiol., 26(2), 111–124. [Google Scholar]
  21. Jo H, Park HR, Choi SJ, Lee SY, Kim SJ, & Joo EY (2021). Effects of Organic Light-Emitting Diodes on Circadian Rhythm and Sleep. Psychiatry Investig, 18(5), 471–477. 10.30773/pi.2020.0348 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Joiner WJ (2019). Neuroscience: sleep fragmentation impairs memory formation. Current Biology, 29(22), R1181–R1184. [DOI] [PubMed] [Google Scholar]
  23. Jonsson BA, Bjornsdottir G, THorgeirsson TE, Ellingsen LM, Walters GB, Gudbjartsson DF, … Ulfarsson MO (2019). Brain age prediction using deep learning uncovers associated sequence variants. Nature Communications, 10, Article 5409. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Kazemi R, Haidarimoghadam R, Motamedzadeh M, Golmohamadi R, Soltanian A, & Zoghipaydar MR (2016). Effects of shift work on cognitive performance, sleep quality, and sleepiness among petrochemical control room operators. Journal of circadian rhythms, 14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Kecklund G, & Axelsson J (2016). Health consequences of shift work and insufficient sleep. Bmj, 355. [DOI] [PubMed] [Google Scholar]
  26. Kim JB, & Kim JH (2017). Regional gray matter changes in shift workers: a voxel-based morphometry study. Sleep Medicine, 30, 185–188. [DOI] [PubMed] [Google Scholar]
  27. Lee K-W, Yang C-C, Chen C-H, Hung C-H, & Chuang H-Y (2023). Shift work is significantly and positively associated with dementia: A meta-analysis study. Frontiers in Public Health, 11, 998464. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Mander BA, Winer JR, & Walker MP (2017). Sleep and human aging. Neuron, 94(1), 19–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Marquie JC, Tucker P, Folkard S, Gentil C, & Ansiau D (2015). Chronic effects of shift work on cognition: findings from the VISAT longitudinal study. Occup Environ Med, 72(4), 258–264. 10.1136/oemed-2013-101993 [DOI] [PubMed] [Google Scholar]
  30. Mirick DK, Bhatti P, Chen C, Nordt F, Stanczyk FZ, & Davis S (2013). Night shift work and levels of 6-sulfatoxymelatonin and cortisol in men. Cancer Epidemiol Biomarkers Prev, 22(6), 1079–1087. 10.1158/1055-9965.EPI-12-1377 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Musiek ES, Xiong DD, & Holtzman DM (2015). Sleep, circadian rhythms, and the pathogenesis of Alzheimer disease. Experimental & molecular medicine, 47(3), e148–e148. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Ning K, Duffy BA, Franklin M, Matloff W, Zhao L, Arzouni N, … Toga AW (2021). Improving brain age estimates with deep learning leads to identification of novel genetic factors associated with brain aging. Neurobiology of Aging, 105, 199–204. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Paixao L, Sikka P, Sun H, Jain A, Hogan J, Thomas R, & Westover MB (2020). Excess brain age in the sleep electroencephalogram predicts reduced life expectancy. Neurobiology of aging, 88, 150–155. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Park C, Bang M, Ahn KJ, Kim WJ, & Shin N (2020). Sleep disturbance-related depressive symptom and brain volume reduction in shift-working nurses. Scientific reports, 10(1), 9100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Park HR, Choi SJ, Jo H, Cho JW, & Joo EY (2020). Effects of Evening Exposure to Light from Organic Light-Emitting Diodes on Melatonin and Sleep. J Clin Neurol, 16(3), 401–407. 10.3988/jcn.2020.16.3.401 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Rahman SA, St. Hilaire MA, & Lockley SW (2017). The effects of spectral tuning of evening ambient light on melatonin suppression, alertness and sleep. Physiology & Behavior, 177, 221–229. https://doi.org/ 10.1016/j.physbeh.2017.05.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Ramesh V, & Gozal D (2009). Sleep fragmentation differentially modifies EEG delta power during slow wave sleep in socially isolated and paired mice. Sleep Science, 2(2), 64–75. [Google Scholar]
  38. Redeker NS, Caruso CC, Hashmi SD, Mullington JM, Grandner M, & Morgenthaler TI (2019). Workplace interventions to promote sleep health and an alert, healthy workforce. Journal of Clinical Sleep Medicine, 15(4), 649–657. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Rouch I, Wild P, Ansiau D, & Marquié J-C (2005). Shiftwork experience, age and cognitive performance. Ergonomics, 48(10), 1282–1293. [DOI] [PubMed] [Google Scholar]
  40. Smith SM, Vidaurre D, Alfaro-Almagro F, Nichols TE, & Miller KL (2019). Estimation of brain age delta from brain imaging. Neuroimage, 200, 528–539. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Sun H, Paixao L, Oliva JT, Goparaju B, Carvalho DZ, Leeuwen K. G. v., … Westover MB (2019). Brain Age from the Electroencephalogram of Sleep. Neurobiol Aging, 74, 112–120. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Torquati L, Mielke GI, Brown WJ, Burton NW, & Kolbe-Alexander TL (2019). Shift work and poor mental health: a meta-analysis of longitudinal studies. American journal of public health, 109(11), e13–e20. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Wickwire EM, Geiger-Brown J, Scharf SM, & Drake CL (2017). Shift work and shift work sleep disorder: clinical and organizational perspectives. Chest, 151(5), 1156–1172. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Wright KP Jr, Bogan RK, & Wyatt JK (2013). Shift work and the assessment and management of shift work disorder (SWD). Sleep medicine reviews, 17(1), 41–54. [DOI] [PubMed] [Google Scholar]
  45. Wyse CA, Celis Morales CA, Graham N, Fan Y, Ward J, Curtis AM, … Biello S (2017). Adverse metabolic and mental health outcomes associated with shiftwork in a population-based study of 277,168 workers in UK biobank. Annals of Medicine, 49(5), 411–420. [DOI] [PubMed] [Google Scholar]
  46. Ye E, Sun H, Leone MJ, Paixao L, Thomas RJ, Lam AD, & Westover MB (2020). Association of sleep electroencephalography-based brain age index with dementia. JAMA network open, 3(9), e2017357–e2017357. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Yook S, Choi SJ, Zang C, Joo EY, & Kim H (2024). Are there effects of light exposure on daytime sleep for rotating shift nurses after night shift?: an EEG power analysis. Frontiers in Neuroscience, 18, 1306070. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Yook S Park HR Park C Park G Lim DC, Kim J, … Kim H (2022). Novel neuroelectrophysiological age index associated with imaging features of brain aging and sleep disorders. Neuroimage, 264, 119753. [DOI] [PubMed] [Google Scholar]
  49. Özdemir PG, Selvi Y, Özkol H, Aydin A, Tülüce Y, Boysan M, & Beşiroğlu L (2013). The influence of shift work on cognitive functions and oxidative stress. Psychiatry research, 210(3), 1219–1225. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS2006016-supplement-1.docx (153.3KB, docx)

RESOURCES