Study design and participants
This longitudinal study was conducted at a university hospital in Japan. The recruitment period for this study was from October 1, 2021, to March 31, 2022, and each participant was followed for 3 months. Participants included 40 female nurses aged 23–41 years working in a two-shift system (two 12-h shifts within 24 h) in a general ward. The gender and age of the participants were limited as factors affecting cortisol levels [29]. The exclusion criteria were as follows:
-
New graduate nurses;
-
Administrators;
-
Nurses who regularly used sleeping pills, antipsychotics, antidepressants, steroids, and oral contraceptives;
-
Pregnant nurses;
-
Nurses on leave; and
-
Nurses undergoing treatment for anemia, thyroid disease, diabetes, menstrual irregularities, insomnia, irregular heartbeat, insomnia, and autonomic nervous system disorders.
Posters outlining this study were distributed by the administrator to nurses in the target wards to recruit participants. This study was approved by the Ethics Review Committee of the Faculty of Health Sciences, Hokkaido University (reference No. 21-43) and was conducted in accordance with the Declaration of Helsinki. All participants provided written informed consent. Participants were offered a gratuity (Quo card) worth 12,000 yen.
Study procedures
Participants completed questionnaires regarding their demographics, working conditions, fatigue, and burnout. During the first month, they were asked to collect saliva samples upon awakening during three different day shifts. After 3 months, they reported the extent of their fatigue, burnout, and reluctance to stay (Fig. 1). In our study, the time interval between assessing predictors and the outcome evaluation was three months, which corresponds to a quarter term. This relatively short-term assessment of turnover risk has the following advantages in terms of nursing shortage: (1) early detection of turnover risk and prevention of problems becoming more severe, and (2) increased opportunity to identify challenges and dissatisfaction among nurses.
Study procedure. Black circles indicate measurements
Considering the several influencing factors [30], the measurement of salivary cortisol at a single time point is not appropriate [29, 31] Two methods are used for multiple measurements: one is taking measurements multiple times on the same day (e.g., awakening response, diurnal variation) [32, 33] and the other is an inter-day trend [29, 34]. Our study considered inter-day trends to eliminate the burden of measurement and the influence of work conditions. Additionally, salivary cortisol measured in the early morning is associated with work stress and fatigue among nurses [23, 29, 31]. Considering the inter-day trend due to differences in work patterns and other factors, three different day shifts (one-time point more than in a previous study) were set as the measurement days [29].
Reluctance to stay in the current job
Reluctance to stay was measured at three months using the question “How reluctant do you feel about continuing on your current job?” Participants responded to a rating scale ranging from 0 (not at all) to 10 (extremely). In related studies, nurses’ intention to leave is often assessed using a single item [14, 35, 36]. Additionally, our primary focus was to examine whether the cortisol profile is a predictor of reluctance to stay. In practice, the detailed background of nurses’ intention to leave can be ascertained through subsequent interviews by administrators after risk screening. Based on the above, we assessed only the extent of reluctance to stay using the aforementioned single question.
Salivary cortisol
Participants gargled upon awakening, rested for 5 min, put on disposable gloves, and collected saliva samples with an oral fluid collector (OFC) swab (SOMA bioscience, Oxfordshire, United Kingdom). The OFC swab containing the saliva was placed into a 3-mL buffer solution. The buffer bottle containing the OFC swab was mixed for 2 min, then the OFC swab was removed from the bottle. Participants were asked in advance to avoid heavy exercise and alcohol consumption the day before saliva collection and to avoid eating, drinking, and brushing their teeth until the saliva was collected. Participants also reported their waking time, physical symptoms, and mood (e.g., irritability, depression) on the day of saliva collection, none of which were significantly correlated with each cortisol level. Saliva sample bottles were sealed in light-shielded bags, maintained at 37 °C or below, and submitted to the researcher upon arrival at work.
The saliva sample bottles were collected from the participants were kept frozen (up to 2 months) until analysis according to the instrument’s manual in order to maintain sample stability. The buffer solution containing saliva was placed in a cortisol lateral flow device (LFD) with three drops. A SOMA CUBE Reader (SOMA bioscience, Oxfordshire, United Kingdom) was placed on the LFD to measure cortisol concentrations. Cortisol levels measured using the SOMA CUBE reader [28, 37] and applying the same principle of measurement [38, 39] have revealed a positive correlation with the enzyme-linked immunosorbent assay (ELISA) method.
Each participant received an individual orientation session (lasting approximately 1 h) during which the detailed method of saliva sampling was explained. In this orientation, they received a manual which indicated the process and fully explained that saliva should be collected following 5 min of rest after awakening. The participants agreed to report any failure to adhere to these instructions to the researchers and understood that sampling dates would need to be rescheduled in such cases. No participant reported any violations of the sampling protocol, confirming their adherence to the procedure.
Burnout
We used the verified and reliable Japanese Burnout Scale [40], developed in accordance with the Maslach Burnout Inventory to evaluate the participants’ burnout across three dimensions: emotional exhaustion, depersonalization, and decline in personal accomplishment. The inventory consists of 17 items on a five-point rating scale, ranging from 1 (never) to 5 (always). The average of the item scores included in each factor was used as the factor score. In each factor, a higher score indicates a stronger state of the condition.
Chronic fatigue
We used the Japanese version of the 15-item Occupational Fatigue/Exhaustion Recovery Scale [41], and only 5 items of chronic fatigue were used in our analysis. The items are constructed on a seven-point Likert scale ranging from 0 (strongly disagree) to 6 (strongly agree). The standardized score of chronic fatigue (range 0–100) was calculated using the following formula: sum of the scores of the five items applicable to chronic fatigue divided by 30 and multiplied by 100. A higher standardized score indicates a stronger extent of chronic fatigue.
Demographic factors and working conditions
A self-administered questionnaire was used to assess participants’ demographic factors and work conditions. The factors included age, job tenure, body mass index (kg/m2), marital status, childcare, family care role, alcohol consumption (drink or not drink), smoking habits (yes or no), and leisure time activities [42]. Working conditions included the number of night shifts worked per month; the total overtime hours per month (< 10 h, < 20 h, ≥ 20 h); the experience of less than 11 h of rest between shifts (a quick return) in the previous month; and perceived change in workload (i.e., decreased, unchanged, or increased vs previous month).
Statistical analysis
The sample size was calculated using the G*Power version 3.1.9.7 (Universität Kiel). For hypothesis testing, we used a linear regression model with “reluctance to stay” as the objective variable and cortisol profile as the predictor, effect size (f2) = 0.20 (medium) [29], significance level = 0.05, power = 0.80, and a sample size of at least 42 would be sufficient.
Data were summarized using means (standard deviations [SD]) or frequencies (percentages). To normalize the distributions, all cortisol data were log-transformed before analysis log10 (X). For ease of interpretation, the results show the untransformed values. Correlations between variables were evaluated by Pearson correlation analysis. There was no consistent correlation between cortisol levels on each day shift and reluctance to stay (day shift 1: r = − 0.450, day shift 2: r = − 0.256, day shift 3: r = − 0.368). Because consistent levels across two different day shifts are associated with chronic fatigue in nurses [29], previous studies [43] have used the average cortisol level over two days. In contrast, this study used more time points for cortisol measurements, with an average of three measurements, to represent the cortisol profile.
Four separate linear regression models were performed with “reluctance to stay” as the objective variable. First, we assessed the association between the cortisol profile and reluctance to stay by a univariate regression model for our hypothesis (Model 1). Next, we adjusted for chronic fatigue and burnout during the enrollment of the study and examined the association of the cortisol profile (Model 2). Following this, a multiple regression model was performed with job tenure significantly associated with “reluctance to stay” (r = − 0.425, P = 0.006) and subjective indicators as explanatory variables (Model 3). The subjective indicators in Model 3, selected based on the lowest Akaike information criterion using the backward elimination method, were chronic fatigue and decline in personal accomplishment. Finally, Model 4, which added the cortisol profile to the explanatory variables, evaluated the association of the cortisol profile. In all models, we reported the coefficient of determination (R2) and the adjusted R2.
Statistical analysis was performed using JMP Pro software, version 16.1 (SAS Institute Inc., Cary, NC, USA), with P < 0.05 considered statistically significant.
Add Comment