Impact of shift work on the cortisol awakening response and stress: a longitudinal study in female shifting monthly nurses

Background: Although the majority of shift nurses are female, there is still an expectation that they fulfil the traditional role of women in the family, often conflicting with shift work, increases stress, and affects cortisol secretion patterns. This study was to understand the changes in the cortisol awakening response (CAR) and work stress in nursing personnel working in different shifts. Methods: 41 female shift nurses were recruited that fit criteria using purposive sampling. We obtained personal information, administered the Taiwan Nurse Stress Checklist (NSC), and the nurses themselves collected saliva samples upon waking and 30 minutes after waking for three consecutive days at home. We then analyzed the data obtained using a hierarchical linear model. Results: The results revealed that night-shift nurses derived significantly more stress from an inability to complete personal tasks than did day-shift nurses ( B = 4.39, p < .001) or evening-shift nurses ( B = 3.95, p < .001). Night-shift nurses also exhibited significantly lower CARi than did day-shift nurses ( B = -3.41, p < .001) or evening-shift nurses ( B = -2.92, p < .01) as well as significantly lower cortisol levels 30 minutes after waking than did day-shift nurses ( B = -3.88, p < .001) or evening-shift nurses ( B = -3.31, p < .01). Conclusions: This study indicate that female night-shift nurses displayed the lowest CARi and cortisol levels 30 minutes after waking and are more negatively affected by being unable to complete personal tasks.


Background
The work environments and duties of nursing personnel are extremely complex.
They must monitor the mental and physical conditions of their patients, meet the needs of the family members of their patients, and communicate with their supervisors, co-workers, and other medical personnel. In addition to their work environments being full of noise and communicable diseases, they must also cope with heavy workloads and time stress. These are all sources of stress in nursing [1].
Moreover, women are still expected to serve the traditional roles in their families, for example, by cooking, doing housework, and caring for children and elders. This career model in which women must meet the demands of both family and work thus places even more stress on female nursing personnel [2].
Researchers have identified that shift work itself is a stressor. Nurses working shifts must continuously adjust to changes in the external environment, which means ever-changing lifestyles and psychological stress [3]. Altering the time of waking can have a profound impact on cortisol levels, which are a crucial activity index of neuroendocrine systems faced with stress [4]. The hypothalamus-pituitary-adrenal (HPA) axis is extremely meaningful in regard to responses to physiological or psychological stimuli of the human body. The balanced operations of this endocrine system rely on a regular sleep schedule, the keeping of which in turn relies on the normal fluctuations in cortisol levels during the last 24 hours. Cortisol levels of an individual are highest upon waking in the morning and lowest at night during sleep.
Only in this way can a regular sleep schedule be maintained [5][6][7]. Moreover, cortisol secretion is a physiological self-defense mechanism to cope with stress [8].
HPA axis stimulation has a stronger impact on hormones when it takes place in the morning than when it happens in the evening [5,9]. Thus, sleep schedules are closely associated with cortisol levels. Vangilova et al. also indicted that working at night results in lower cortisol levels in the morning after, but after the fifth night of working at night, cortisol levels will become the inverse of the original day-night cycle [10].
The cortisol awakening response (CAR) plays a crucial role in human HPA axis activation during the awakening process. It is an important indicator of HPA axis activity and is highly associated with the arousal system in the brainstem [11]. CAR is widely accepted as being fairly stable; cortisol levels in the human body swiftly rise upon wakening and peaks 30 minutes later. At present, it is known to be associated with energy increase or anticipated stress, which elevates average cortisol levels by 50-100%. Such levels persist for at least an hour and then slowly decline. Subsequently, secretions decline from day to night and maintain periodic changes [12,13].
Although an increasing number of studies are using CAR to detect adrenocortical activity, most studies on nursing personnel working shifts examined stress and CAR on a single level. No relevant studies have used multilevel models so far. The hierarchical linear model (HLM) is currently one of the most suitable statistical methods for multilevel data analysis. The model regards observation time points as the first level and the basic attributes of nursing personnel as the second level while examining the relationships among variables. We thus performed a longitudinal analysis on the panel data collected and used the HLM growth model to (1) understand the changes in CAR and work stress in nursing personnel working in different shifts, (2) determine the influence of specific attributes of female nurses (age, religion, marital status, educational background, and years of service) on CAR and work stress, and (3) examine the relationship between CAR and stress. Figure 1 displays the conceptual framework of this study.

Study design
This study investigated nurses working shifts at a teaching hospital in Northern Taiwan. This study investigated nurses working shifts at a teaching hospital in Northern Taiwan. A portion of the data collection has already been published [14].
We recruited 41 nurses from January of 2017 to January of 2018 who were working monthly rotating shifts between the ages of 20 and 45 with more than one year of experience as a clinical nurse. Nurses who were pregnant or were using contraceptives or hormone medication were excluded.
Repeated measurements were performed on shift nurses at a teaching hospital in Northern Taiwan to determine the influence of different shifts and work stress on CAR. Wirth et al. found that individuals who sometimes work night shift or evening shift wake up with lower CAR values than those who permanently work day shift [16]. The difference between these two groups of individuals is greatest on the 5th day after a shift change and gradually decreases from the 7th day to the 14th day after a shift change. To eliminate the influence of shift changes, we conducted the first structured interviews 14 days after shift changes. Subsequent structured interviews were then conducted on the 14th day of the next month and the 14th day of the following month for follow-up data collection. Our procedure follows nurses across three consecutive shift changes over three months that cover day, evening, and night shifts.

Research tools
The measurement tools included a basic information sheet, the Taiwan Nurse Stress Checklist (NSC), and a radioimmunoassay (RIA): Basic information sheet: This included information on the age, religion, marital status, educational background, and years of service of the nurses.
Taiwan Nurse Stress Checklist (NSC): The Taiwan NSC is based on the scale developed by Benoliel to gauge the work stress that a nurse has experienced during the week prior [17]. Each question item was measured using a nine-point Likert scale ranging from 0 to 8 points, with a higher score indicating greater stress. The scale was divided into five domains: personal reactions, personal concerns, work concerns, role competence, and work completion concerns. Internal consistency ranged from 0.80 to 0.91. Tsai and Chen revised the scale for application in Taiwan shopping. The Cronbach´s α of each aspect was greater than 0.84. cortisol in the saliva of the participants. Measuring the physiological indicators of saliva is a noninvasive approach to measure stress which is simple and costeffective. At present, while it is known that cortisol levels in saliva is a tenth of those in serum, the variations in the two are the same, so changes in cortisol levels in saliva accurately reflect changes in serum cortisol levels. A single saliva sample presents inadequate stability in cortisol results, so saliva samples were collected for three consecutive days at two different times to measure CAR variations, one immediately upon waking and the other at 30 minutes after waking [12,19]. Many studies use the same two timepoints to measure CAR [20,21]. CAR plays a crucial role in increasing the energy expected to be necessary for the human body to cope with a day's work [22]. CARi (the increase in cortisol following waking) is an extremely valid and standardized method of measuring HPA axis responsiveness and estimates the average increase in cortisol levels during the first 30 minutes after waking [23]. For the sake of stability in the cortisol saliva sample data, the participants were asked not to eat within the 30 minutes before taking their sample and collect more than 3 ml of saliva. All saliva samples had to be frozen for storage after being taken. Prior to analysis, the samples were stored in a freezer at -20℃.
For analysis, the samples were thawed, mixed, and then centrifuged to remove particulate matter. System precision tests on the cortisol analysis system were performed five times as suggested by Wilson and Miles [24]. The intra-assay CV % and inter-assay CV % of the cortisol RIA analysis system were 0.96% and 12.09%, respectively, and the detection limit was 1.66 nmol/l.

Statistical Analyses
We first employed SPSS for Windows 21.0 (SPSS, Chicago, IL, USA) to obtain the descriptive statistics and subsequently performed an ANOVA. Using HLM 6.03 (SSI, 2010), we then conducted HLM analysis on the panel data obtained. In model selection, we first used the null model to derive within group variance (σ2) and between group variance (τ00). According to Cohen, an intraclass correlation coefficient (ICC), which equals (τ00)/(τ00 + σ2), greater than 5.9% (moderate effects) indicates significant variance in the result variables of individual participants [25]. Using a random coefficients regression model and an interceptsas-outcomes model, we then determined whether the basic attributes and shifts of the nurses exert significant influence on stress and CAR.

Basic attributes of female shift nurses
The mean age of the participating nurses was 25.0 (range = 21.0-40.0). In terms of religion, the largest group (48.8%) was not religious. The vast majority of the participants (95.1%) were either single or divorced. Most (70.7%) had a bachelor's degree or higher. With regard to years of service, the largest group (43.9%) have worked for 1-3 years (Table 1).  25) . This means that the interclass differences among the nurses of different shifts in CARi, cortisol levels upon waking, and cortisol levels 30 minutes after waking cannot be ignored and that cross-level analysis is necessary [26].
Thus, a respective 32.08%, 53.79%, and 49.52% of the total variance in CARi, cortisol levels upon waking, and cortisol levels 30 minutes after waking are caused by interindividual variance.
As shown in Table 2  As can be seen in Table 3  Influence of work stress on CAR An ANOVA of the stress from personal reactions, work concerns, work competence, and inability to complete personal tasks for nurses working the three different shifts presented no significant differences. With the shift and other personal attributes controlled, the four aspects of work stress also showed no significant differences in their influence on the CARi, cortisol levels upon waking, or cortisol levels 30 minutes after waking of female shift nurses (Table 3).

Discussion
Waking time and stress are both considered factors influencing CAR, but the correlations with these factors are still unclear. This study is the first to adopt a longitudinal approach to examine the differences among women working different shifts (and waking up at different times) and their correlations with CAR and work stress. The study results revealed that after the personal attributes of the nurses were controlled, night-shift nurses displayed significantly lower CARi and cortisol levels 30 minutes after waking than did day-shift nurses or evening-shift nurses.
When religion, marital status, educational background, and years of service were controlled, the results showed that older nurses derived less stress from work concerns. Following the control of the personal attributes of the nurses, the results indicated that the night-shift nurses derived more stress from the inability to complete personal tasks than did day-shift nurses or evening-shift nurses.
Cortisol plays a crucial role in regulating the coping mechanism of chronic stress.
Shift work exerts influence on multiple physiological, neuroendocrine, and hormonal functions. As a result, CAR is currently considered to be a major index of shift work tolerance [27]. Research has shown that changes in CAR may be the result of increased stress and impaired sleep quality following shift changes; however, controlling for shift factor revealed no significant differences in CAR variations [28].
Many existing studies presented different findings. Federenko  likely due to the normal circadian rhythm of the latter group [31,32].
The results of this study also revealed that work stress which shift nurses must cope with varies with age, with older shift nurses feeling less stress from work concerns.
Nursing personnel work on the front lines and must have good professional decisionmaking and execution abilities, be skilled, and have good relationships with patients and other medical personnel, all of which come with increased work experience, because they come into direct contact with patients and their family members.
Thus, young nurses must invest more energy in their work and perhaps this causes them to perceive more work stress.  [37]. Thus, only moderate and severe amounts of stress and poor selfperceived health alter CAR, which means that a low level of stress does not affect CAR or self-perceived health. This is consistent with the results of our study. After the personal attributes and shifts of the nurses were controlled, we found that the various aspects of work stress did not significantly affect CARi. One explanation for this observation is that the month-long tracking periods of the participants may not be long enough for any significant changes to appear in the various aspects of work stress, so indeed, CARi would not be affected.

Conclusions
The results of this study indicated that female night-shift nurses displayed the lowest CARi and cortisol levels 30 minutes after waking and were more troubled with the inability to complete personal tasks. Our results imply that monitoring cortisol levels may be necessary to understanding the HPA axis variations in nurses that work shifts. These results also provide hospital managers with a reference when arranging nurse shift schedules. Suitable shift arrangements could reduce the impact of shift work on female nurses and their health.

Limitations
The number of samples collected in this study was limited and we only investigated female nurses at a single hospital, so the results may not be applicable to other hospitals, other departments, or male nurses. CAR changes were based on only two timepoints: upon waking and 30 minutes after waking. We suggest that future studies include three or more timepoints to measure the changes in CAR which could lead to a more comprehensive analysis.

Ethics approval and consent to participate
With the approval of the institutional review board (IRB), we explained our objectives and procedure to the research participants and gained their consent. The approval number from the Office of Human Research, Taipei Medical University is N201703064.

Consent for publication
Not applicable.

Availability of data and materials
The datasets generated and/or analyzed during the current study are not publicly available due to considerations of data protection.

Competing interests
The authors have no conflict of interest to disclose.

Funding
Research reported in this publication was supported by the Taipei Medical  Mean cortisol awakening responses and standard deviation of 41 female nurses working diffe