Association Between Shift Working and Brain Morphometric Changes in Workers: A Voxel-wise Comparison

Abstract

Objective

There is abundant evidence from observational studies linking various health problems to shift work, but there is a lack of brain-based neurological evidence. Therefore, we examined morphometric changes on brain magnetic resonance imaging (MRI) between shift and non-shift workers.

Methods

A total 111 healthy workers participated in this study and underwent brain MRI, with the analysis incorporating merged workers' health surveillance data from regional hospital workers. Voxel-based morphometry analysis was used to investigate regional changes in the gray matter volume. To investigate the association of structural changes between shift workers and non-shift workers, a general linear model and threshold-free cluster enhancement were used with covariates, including total intracranial volume, age, and sex.

Results

After family-wise error correction, non-shift workers exhibited a significantly larger cerebellar region (p < 0.05) than shift workers. Conversely, the inferior parietal gyrus was found to be significantly larger in shift workers than in non-shift workers with family-wise error correction.

Conclusions

We observed increased clusters in the brains of both shift and non-shift workers, suggesting that the acquired occupational environment, including the shift work schedule, could influence brain neuroplasticity, which is an important consideration for occupational health.

Key Messages.

Shift work has a negative impact on the health of workers. Although, most of the health problems from shift work are related with circadian rhythm from brain, there is a lack of brain-based neurological evidence. We examined morphometric changes on brain between shift and non-shift workers.

Non-shift workers showed a significantly increased cerebellar region (center for movement) than shift workers. The inferior parietal gyrus (center for communication) was found to be significantly larger in shift workers than in non-shift workers.

These findings suggest that preclinical brain neurologic changes might be affected from occupational conditions such as working schedule; it could be a cornerstone for research of workers health.

1. Introduction

Shift work has a negative impact on the health of workers [1]. Compared with non-shift workers, shift workers are more likely to have gastritis, breast cancer, and mental problems (e.g., sleep disorders, depression, and suicide) and are more at risk of occupational accidents or injuries [[2], [3], [4]]. Most health effects associated with shift work are explained by circadian rhythm impairment, which is caused by disruption of the neurological pathway from the optic nerve to melatonin [5]. Although there is abundant evidence from observational studies linking various health problems to shift work, there is a lack of brain-based neurological evidence.

Because a clear biological and neuroscientific basis for shift work–related diseases has not been elucidated, brain-based neurological studies can help contribute to our understanding with existing evidence. Steps have been taken toward understanding the neuromechanisms in occupational medicine; these studies have primarily concentrated on hazardous substance exposure and occupational patterns [6,7]. However, there is a notable knowledge gap concerning occupation-induced dysfunction or diseases. Furthermore, there have been insufficient investigations into the health effects of shift work, which is the focus of our research. Shift work is also considered a major occupational risk factor, and among its health effects, mental problems and sleep disturbances have been associated with clinical changes in brain structure in the previous studies [8,9]. However, there are only a few preclinical neuroscientific studies on brain structural changes as health-related effects of shift work. A recent study by Bittner et al. [10] conducted a multimodal investigation into the association between shift work and brain structure in a population-based research, finding no significant differences attributable to shift work. A study of nurses found an association between brain structures linked to sleep disturbances and depressive symptoms and shift work. [11]. Existing studies have attempted surface-based analyses of the relationship between shift work and the brain, focusing on cortical thickness and surface area, or have approached the study by examining differences in specific brain regions based on narrow hypotheses [10,11]. While surface-based analyses are appropriate for exploring changes in individuals with time, voxel-based morphometry (VBM) is more appropriate for comparing between groups. VBM allows for whole-brain exploratory analyses without requiring prior hypotheses about specific areas.

The purpose of this study is to examine the health effects of shift work in a preliminary way by comparing brain structure. Therefore, in our study, we conducted a VBM-based exploratory analysis of the changes in brain structure associated with shift work among healthcare workers who had normal brain function without neurological or clinical abnormalities.

2. Materials and methods

2.1. Ethics statements

Studies involving human participants were reviewed and approved by the Institutional Review Board (IRB) of Gachon University Gil Medical Center (IRB number: GCIRB2021-150). All the participants provided written informed consent to participate in this study.

2.2. Study design and population

This study used merged data from the Gachon Regional Occupational Cohort Study and MRI from research on the effect of shift work on brain structure, supported by the National Research Foundation of Korea. Gachon Regional Occupational Cohort Study has been a large-scale longitudinal study on occupational safety and health since 2018 [12]. It includes data on work environment monitoring, worker health surveillance, and occupational health services from approximately 200 companies and 35,000 workers. We selected one medical facility that was a regionally representative tertiary hospital, and healthcare workers were encouraged to participate via further trained interviewers who conducted the survey and performed brain MRI. The additional survey included information about the type of work, educational level, job duration, working schedule, and chronic diseases, or neurological conditions related to the brain structure. We conducted further evaluations of 115 healthcare workers from May 2022 to August 2023. The MRI scans were not measured repeatedly, so cross-sectional study methods and interpretation apply to this study. The current study included 111 participants, excluding four workers who were on maternity leave or had low-quality MRI.

2.3. Variables

The baseline characteristics included sex, age, and educational level. The dominant hand was also recorded to interpret the brain images. Work-related factors included job duration and work type (e.g., nurse, physician, and others). Others included radiologists, pathologists, or other hospital workers. Shift work was defined by survey responses about work schedules used in the Korean Working Condition Survey [13]. The survey asked about current work schedule, of which any work schedule other than fixed daytime work was defined as shift work. Additionally, we asked when they started shift work to calculate the duration of shift work.

2.4. Data acquisition and processing

A three-dimensional T1-weighted magnetization prepared rapid gradient echo (T1w MPRAGE) sequence that was acquired using the following parameters: repetition time, 1970 ms; echo time, 2.84 ms; inversion time, 991 ms; field of view, 256 × 256; flip angle, 9; in-plane resolution, 0.5 × 0.5 × 1 mm³; number of slices, 192, and scan time, 4 minutes 34 seconds.

The acquired T1w MPRAGE images were used to generate VBM maps to investigate regional volumetric changes in the brain [14]. As shown in Fig. 1, we performed segmentation, normalization, data quality checks, and smoothing for preprocessing of the VBM using the computation anatomy toolbox (CAT)12 in statistical parametric mapping [14], which divides the brain into gray matter, white matter, cerebrospinal fluid, and skull, using a Gaussian mixture model [14]. VBM analysis was conducted within the gray matter to observe changes in gray matter volume. After brain segmentation, the segmented gray matter images were spatially normalized to a T1-weighted MNI 152 template image using nonlinear registration [15]. This step was necessary because of individual variations in brain structures when comparing brain images between the shift and non-shift workers. For the third preprocessing step, we conducted a data quality check, including a check for sample homogeneity based on the interquartile range in CAT 12. Because subsequent statistical analyses can be influenced by the preprocessing steps, it is essential to verify whether the normalization and segmentation processes have been performed effectively. Finally, Gaussian smoothing with an 8-mm full width at half maximum was applied to reduce noise, improve registration, and enhance statistical validity [14].

Fig. 1.

MRI data processing and voxel-wise statistical analysis of the differences between shift workers and non-shift workers. MRI, magnetic resonance imaging.

2.5. Statistical analysis

Differences in baseline characteristics according to the work schedule were calculated using the Chi-square test or t test. To investigate the structural changes associated with shift and non-shift work, a general linear model (GLM) based on two-sample t tests was created to compare the VBM between the two groups using CAT12. In the GLM, total intracranial volume, age, and sex were used as covariates. Considering the inherent individual differences in brain volume, we included total intracranial volume as an independent variable. Since age and gender can influence VBM, we used these variables in GLM as independent variables. Whole-brain results from the GLM were evaluated using threshold-free cluster enhancement (TFCE) inference for multiple comparison correction [16]. Both uncorrected and family-wise error (FWE)-corrected thresholds (p < 0.05) were examined. Combining the GLM with TFCE has been recommended as the optimal configuration for advanced VBM analyses (Radua, 2014).

3. Results

The baseline characteristics of the study participants (32 shift workers and 79 non-shift workers) are presented in Table 1. Differences in sex, dominant hand, and type of work, according to shift work, were not significant. The shift workers were younger and more educated than the non-shift workers. However, most of the participants were younger than 45 years of age (fairly free of age effects of neurological function) and highly educated (all participants had graduated from college or higher). Job duration was longer among non-shift workers than shift workers. This is because most hospital workers tended to be removed from shift work schedules as they gained experience.

Table 1.

Baseline characteristics of study participants according to shift work

Characteristic	Total (n = 111)	Shift work		Total (n = 111)
		No (n = 79)	Yes (n = 32)
Sex
Male	40 (36.0)	28 (35.4)	40 (36.0)	0.8387
Female	71 (64.0)	51 (64.6)	71 (64.0)
Age (years)	36.11 ± 9.63	38.49 ± 9.74	36.11 ± 9.63	0.0001
Education level
College or university	40 (36.1)	33 (41.8)	40 (36.1)	0.0490
Graduate school or higher	71 (63.9)	46 (58.2)	71 (63.9)
Dominant hand				0.1481
Right	108 (97.3)	78 (98.7)	108 (97.3)
Left or both	3 (2.7)	1 (1.3)	3 (2.7)
Job duration (years)	(8.31 ± 8.62)	(9.82 ± 9.43)	(8.31 ± 8.62)	0.0002
Type of work				0.9796
Others	46 (41.5)	33 (41.8)	46 (41.5)
Nurse	36 (32.4)	25 (31.6)	36 (32.4)
Physician	29 (26.1)	21 (26.6)	29 (26.1)

Data are presented as n, %, or mean ± standard deviation.

Table 2, Table 3 show significant clusters (>500 mm³) using TFCE inference to compare different conditions: non-shift work > shift work or non-shift work < shift work (p < 0.001, uncorrected). When comparing the brain volume in the non-shift work group with that in the shift work group, the regions of the cerebellum, calcarine, cuneus, and lingual gyrus were significantly larger in non-shift workers than in the shift workers. However, upon applying multiple comparison correction to achieve a p < 0.05 (FWE-corrected), it became evident that only the cerebellar region remained statistically significant (Fig. 2, left column).

Table 2.

Significant clusters, peak coordinates, and brain locations following TFCE inference for a contrast where the volume of non-shift work is larger than that of shift work

Definition of the difference	Cluster (mm3)	Peak coordinate (mm)	Peak VBM values		Brain location
			Shift work	Non-shift work
p < 0.001∗	21,504	-20, -41, -60	0.170 ± 0.023	0.183 ± 0.029	Left cerebellum
	13,833	22, -40, -57	0.195 ± 0.025	0.211 ± 0.033	Right cerebellum
	6,975	2, -92, 30	0.185 ± 0.034	0.187 ± 0.030	Left/right calcarine, left/right cuneus, left/right lingual
p < 0.05†	276	-22, -42, -45	0.312 ± 0.042	0.337 ± 0.051	Left cerebellum

The brain locations follow the automated anatomical labeling atlas.

TFCE, threshold-free cluster enhancement; VBM, voxel-based morphometry.

^∗uncorrected p value with a cluster size larger than 500 mm³.

^†Family-wise error corrected p value.

Table 3.

Significant clusters, peak coordinates, and brain locations following TFCE inference for a contrast where the volume of shift work is larger than that of non-shift work

Definition of the difference	Cluster (mm3)	Peak coordinate (mm)	Peak VBM values		Brain location
			Shift work	Non-shift work
p < 0.001∗	6,085	37, 64, 9	0.235 ± 0.033	0.2106 ± 0.032	Right middle/superior frontal, right middle/superior frontal
	2,153	-49, -45, 41	0.345 ± 0.066	0.2879 ± 0.052	Left inferior parietal
	2,033	-15, 54, 3	0.223 ± 0.045	0.1941 ± 0.037	Left superior frontal, left superior/medial/middle frontal, left anterior cingulate
	1,145	-13, 71, 13	0.290 ± 0.042	0.257 ± 0.039	Left superior/medial frontal
	1,278	12, 59, -27	0.204 ± 0.030	0.188 ± 0.034	Right superior frontal, right rectus
	1,413	48, 23, 15	0.287 ± 0.058	0.238 ± 0.059	Right triangular/opercular inferior frontal
	2,979	27, -61, 18	0.231 ± 0.054	0.199 ± 0.045	Right cuneus, right precuneus, right calcarine, right superior occipital
p < 0.05†	61	-50, -44, 42	0.359 ± 0.067	0.301 ± 0.052	Left inferior parietal

The brain locations follow automated anatomical labeling atlas.

TFCE, threshold-free cluster enhancement; VBM, voxel-based morphometry.

^∗uncorrected p value with a cluster size larger than 500 mm³.

^†Family-wise error corrected p value.

Fig. 2.

Results of significant clusters using TFCE inference for a contrast to compare different conditions: volume in non-shift worker > volume in shift worker (left) and volume in non-shift worker < volume in shift workers (right). TFCE, threshold-free cluster enhancement.

When examining regions where the volume of shift work was significantly larger than that of the non-shift work, the frontal regions were found to be significantly larger (Table 3) (p < 0.001, uncorrected). However, after applying a multiple comparison correction to achieve a p < 0.05 (FWE-corrected), only the inferior parietal gyrus remained statistically significant (Fig. 2, right column). The detailed brain locations and cluster are summarized in Table 2, Table 3, respectively.

4. Discussion

This study investigated brain morphometric changes related to shift work schedules. Previous studies have indicated that changes in brain structure might be caused by acquired human activity or the environment, such as juggling and one’s occupation, and our results match those observed in earlier studies [17,18]. As this study aimed to examine changes in normal brain structure without clinical neurological abnormalities, there were many difficulties in clearly identifying the differences in brain structure between shift and non-shift workers. However, this study supports the hypothesis that an acquired occupational environment can influence brain neuroplasticity, which may be an important milestone for occupational health.

None of the shift workers showed significantly increased clusters in the left cerebellum compared with the shift workers. The probable mechanism linking cerebellar preclinical structural changes to shift work involves complex neuroplasticity processes influenced by environmental and physiological factors. Neuroplasticity refers to the brain’s ability to adapt structurally and functionally in response to environmental stimuli or stressors. In the context of shift work, chronic sleep deprivation and circadian rhythm disruptions likely play a significant role. Sleep deprivation has been shown to impair neurogenesis and synaptic plasticity, particularly in regions related to motor control and cognitive processing, including the cerebellum [19,20]. Circadian rhythm disruptions, caused by irregular working hours, affect the timing of hormonal release, neuronal activity, and cellular repair processes, potentially leading to structural changes in the cerebellum and other brain regions [21,22]. These disruptions may induce compensatory mechanisms to mitigate the functional demands of shift work during physically or mentally demanding tasks. Such compensatory processes could explain the observed structural differences as the cerebellum plays a central role in motor learning and adaptive behaviors [22].

The cerebrum participates in higher levels of thinking and action [23], whereas the cerebellum performs several functions related to movement and coordination, including maintaining balance, coordinating movement, vision, and motor learning (with some role in thinking, including processing language, and mood). In severe neurological cases, because of the close relationship between the cerebellum and movement, the most common signs of cerebellar disorders involve a disturbance in muscle control, e.g., a lack of muscle control and coordination, difficulties with walking and mobility, slurred speech or difficulty speaking, abnormal eye movements, and headaches [24]. Our study was conducted in workers who were neurologically healthy, so these results must be expanded upon using preclinical or preliminary approaches. Although further research is needed to determine whether balance and movement may be affected by structural differences in the cerebellum of shift workers, observational studies have reported reduced balance and movement in shift workers [25]. A higher risk of musculoskeletal disorders is observed among shift workers than among non-shift workers [26]. Chronic degenerative diseases, such as lower back pain and arthritis, are prevalent; however, accidental injuries are also associated with a higher risk [27]. Moreover, the risk of occupational injuries is higher than expected despite the absence of more musculoskeletal strain or only a small increase in musculoskeletal burden reported in night or shift workplaces [28,29]. Previous observational studies have explained the increased occupational injury risk among shift workers on the basis of the hypothesis that it is a result of decreased attention and ignoring warning signs [25,[30], [31], [32]]. The current findings suggest that relative cerebellar preclinical function may provide scientific evidence of a high risk of occupational injuries in shift workers.

Shift workers showed significantly increased clusters in the left inferior parietal lobe compared with non-shift workers. The inferior parietal lobe is a key neural substrate underlying diverse mental processes, from basic attention to language and social cognition, which define human interaction [33]. We expected shift work to have only negative neurological effects; however, brain areas associated with the main work performance (patient care and related tasks) were enhanced. Shift healthcare workers who are required to work night shifts perform increased patient care duties with lesser staffing during the day [34]. Thus, it can be assumed that night shift workers require enhanced judgment and decision-making skills to perform their jobs, which would promote the enhancement of these brain regions. This suggests that stimulation from the acquired job demands may have contributed to the strengthening of some of the brain structures.

The current results provide important neuroimaging perspectives. First, it revealed changes in the brain related to the occupational environment. Recent neuroplasticity research has focused not only on functional changes, but also on structural changes in the brain resulting from the environment or training. Using VBM, early studies have illustrated brain changes in individuals with prolonged exposure to occupational characteristics [35]. However, recent studies demonstrated short-term structural plasticity after only a few weeks of training [36]. From a neuroplasticity standpoint, our results can be considered significant, demonstrating changes in the brain associated with the occupational environment of shift work rather than a disease.

Second, the statistical techniques used in this study enhanced reliability. This study employed TFCE to enhance the stability and reliability of the results. TFCE improves objectivity and consistency and effectively identifies small patterns by strengthening cluster formation [16]. Additionally, to mitigate the potential for statistical significance by chance in multiple comparisons, this study introduced FWE correction, revealing significant results in the cerebellum and inferior parietal lobe. This suggests a high level of confidence in our results.

VBM is a neuroimaging analysis technique that allows investigation of focal differences in brain anatomy, using the statistical analysis of voxel-wise gray matter concentration or volume differences between groups or correlations with behavioral measures [14]. In our study, we were able to identify regions of brain structure that differed between shift and non-shift workers. Areas such as the frontal regions showed an enhanced signal in shift workers, while cerebellar regions showed an enhanced signal in non-shift workers. These signal differences in specific brain regions can be explained by neurodevelopmental differences, neuroplasticity and learning, compensatory mechanism, genetic or environmental influences, and others [37]. Neuroplasticity and learning or compensatory mechanism is a reasonable explanation as the study involved neurologically healthy participants [38]. Further research is needed to investigate whether differences in specific brain structures between shift workers and non-shift workers might explain the adverse health effects of shift work.

To our knowledge, this is the first study to describe preclinical brain structural changes related to the working schedules of Korean healthcare workers. Nonetheless, this study had limitations owing to the nature of the study design. Brain morphometric changes from voxel-wise comparisons are a computational approach to neuroanatomy that measure differences in the local concentrations of brain tissue [14]. Therefore, it may be premature to interpret the findings directly in terms of occupational health and medicine. However, a previous study has indicated that brain morphometric changes from a voxel-wise comparison showed a correlation between image properties and tissue microstructures, including astrocytes, axons, and blood vessels [39]. Finally, this study focused on healthy workers without neurological symptoms, making it difficult to interpret the clinical implications of the observed structural changes. This limits the direct applicability of the findings to occupational health and medicine. The study did not account for potential confounding factors, such as lifestyle variables (e.g., physical activity, sleep quality, and diet) or psychosocial stressors, which could influence brain structure. The cross-sectional design precludes causal inference, leaving open the question of whether shift work induces these structural changes or whether pre-existing brain characteristics influence adaptation to shift work. Future research should address these limitations by including more diverse populations, controlling for potential confounders, and employing longitudinal designs to explore the temporal relationships between occupational exposures and brain structure. Additionally, advanced neuroimaging techniques and investigations into functional outcomes will be essential to translate these findings into practical implications for occupational health interventions.

This study aimed to obtain neuroscientific evidence explaining several unhealthy conditions in shift workers. However, we observed a statistically significant increase in different clusters of brain regions in shift and non-shift workers. These findings suggest that the occupational environment can influence brain neuroplasticity, thereby providing important evidence for occupational health. Further research on the level of connectivity, blood supply, and activation of synapses as well as simple brain areas will provide scientific evidence for the potential health effects of shift work.

CRediT authorship contribution statement

Joon Yul Choi: Writing – original draft, Visualization, Software, Methodology, Formal analysis, Data curation. Sungmin Kim: Writing – original draft, Validation, Methodology, Investigation, Formal analysis. Yongho Lee: Writing – original draft, Validation, Software, Methodology, Investigation. Dohyeon Kim: Writing – original draft, Methodology, Investigation, Formal analysis, Data curation. Wanhyung Lee: Writing – original draft, Supervision, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization.

Data availability statement

Not available.

Ethics statement

Studies involving human participants were reviewed and approved by the Institutional Review Board (IRB) of Gachon University Gil Medical Center (IRB number: GCIRB2021-150). All the participants provided written informed consent to participate in this study.

Funding

This study was supported by the Ministry of Education of the Republic of Korea and the National Research Foundation of Korea (NRF-2021R1C1C1008871). The funder had no role in the direction or methodology of the study.

Conflicts of interest

All authors have no conflicts of interest to declare.