The study protocol was conducted in accordance with the Declaration of Helsinki and approved by the ethical committee of the Shimane University Faculty of Medicine (#20191106-1). Written informed consent was obtained from all the participants in this study.

The primary endpoint in this study was the changes in endothelial function after overnight duty. The secondary endpoint was the relationship between endothelial function and fatigue or sleep. The intervention was duty work, where the participants may sleep for some time depending on the work. A total of six RH-PAT examinations of vascular endothelial function for each subject, three inspections before daytime work, and three inspections after nighttime work, were independently performed in the morning before breakfast. To measure the degree of fatigue, the Chalder Fatigue Question (CFQ) scale and the Visual Analog Scale (VAS) were used, and the participants’ sleep duration on the night before the RH-PAT examination was recorded each time.

We recruited healthy volunteers over the age of 20 years who had nighttime duty from Shimane University Hospital as experimental subjects. Thirteen healthy hospital workers (6 men and 7 women) with an average age of 31.6 ± 8.6 years and BMI of 20.8 ± 1.7 kg/m², free from medication and without any former history of CVD, hypertension, or diabetes mellitus were registered to this study.

Endothelial function and HRV were estimated using the Endo-PAT device (Itamar Medical Ltd., Caesarea, Israel), which is a non-invasive technology that can record the pulse wave amplitude of the finger arteries with a pneumatic probe to capture pulsatile plethysmograph and detect peripheral arterial tone. Peripheral arterial tonometry (PAT) was performed in response to reactive hyperemia. In the 12 h before the PAT, subjects refrained from smoking, drinking caffeinated beverages, and eating food. In the examination, the subject was in a quiet room with a constant temperature of 20–24 °C and in a comfortable sitting position with their hands at the heart level. A finger probe was placed on the index finger of each hand, and the pressure cuff was placed on the upper arm. The RH protocol included a 7-min baseline measurement, and then the blood pressure cuff on the test arm was inflated to 60 mmHg above the baseline systolic pressure or at least 200 mmHg for 5 min. The occlusion of pulsatile arterial flow was confirmed by the reduction of the PAT tracing to zero. After 5 min, the cuff was deflated, and the PAT tracing was recorded for an additional 5 min. A computer algorithm that automatically normalizes the baseline signal was used to calculate the ratio of the PAT signal to the baseline after the cuff was released and to the contralateral arm. The PAT data were analyzed using a computer in a manner independent of the operator. PAT was also configured to reduce arterial wall tension and increase the range of arterial wall motion without causing potentially confusing vasomotor changes. A value of < 1.67 was considered to indicate endothelial dysfunction, which was determined in other literature in the population at risk of ischemic heart disease [21, 22]. Therefore, in this study, we assigned subjects with RHI < 1.67 to the abnormal group, and those with 1.67 ≤ RHI > 2.07 to the critical value group, and those with RHI ≥ 2.07 to the normal group. The reproducibility of the PAT measurements was verified using statistical procedures reported previously [23]. Endo-PAT provides the augmentation index (AI) and the baseline pulse amplitude. AI (%) adjusted at a heart rate of 75 bpm (AI@75 bpm) was also assessed in this study.

HRV results were available in the time domain and frequency domain formats. The time domain assessed the R-R interval, the standard deviation of normal-to-normal intervals (SDNN), the square root of the mean squared differences of successive normal-to-normal intervals (rMSSD), the number of N–N intervals differing by more than 50 ms (NN50), the NN50 divided by the total number of N–N intervals (pNN50), and the HRV triangular index. The rMSSD reflects short-term heart rate variability, whereas SDNN is attributed to long-term changes in heart rate variability. The SDNN represents the sympathetic and parasympathetic nerve activity, but it does not allow us to distinguish when changes in HRV are due to an increase in sympathetic tone or the withdrawal of vagal tone. The rMSSD and pNN50 indices represent parasympathetic nerve activity. In the frequency domain, data were analyzed for low frequency (LF), high frequency (HF), and the ratio between low-frequency and high-frequency (LF/HF). HF (0.15–0.4 Hz) is related to parasympathetic nerve activity, and LF (0.04–0.15 Hz) is related to sympathetic and parasympathetic nerve activities. The LF/HF ratio represents the sympathetic-vagal balance [24].

The CFQ scale was used to assess fatigue. The scale was developed as a brief assessment tool for primary and secondary care. The version we used includes 11 items that consist of questions about fatigue symptoms, such as tiredness, sleepiness, lack of energy, insufficient muscle strength, difficulty in concentration, and memory. Each item required participants to rate the frequency of symptoms by choosing among four options: “less than usual,” “no more than usual,” “more than usual,” and “much more than usual.” A score ranging from 0 to 3 was given with the use of the Likert scale. The total score of the scale was obtained by adding the rating for each item, ranging from 0 to 33. Alternatively, a bimodal score ranging from 0 to 11 can be used where response options “less than usual” and “no more than usual” are given scores of 0, and “more than usual” and “much more than usual” are given scores of 1. Our study defined 0–11 points as less fatigue, 12–22 as moderate fatigue, and ≥ 23 as severe fatigue. The VAS scale was recorded for each examination based on the degree of subjective fatigue.

All data were presented as the mean ± standard deviation (SD), or frequency and percentage. For parametric data with a normal distribution, Student’s t-tests for paired observations were used as appropriate. Categorical data were evaluated using ANOVA and Fisher’s least significant difference test. Associations between RHI, sleep, and fatigue were examined using simple linear regression. The association between RHI and VAS scores was examined by multiple linear regression analysis to adjust for SBP, SDNP, rMSSD, and HR. Binary logistic regression models were used to explore the association between the RHI (dependent variable) and variates (independent variables) and expressed as odds ratios (ORs) and 95% confidence intervals (CIs). Statistical significance was set at p < 0.05. All statistical analyses were conducted using SPSS (version 25.0; IBM Corp., Armonk, NY, USA).

This study was conducted with 13 healthy participants, and the data were collected for statistical analysis. A total of 72 RH-PAT examinations were completed except for six other examinations. The RHI, CFQ, VAS scale, sleep duration, heart rate (HR), and HRV of the study participants were evaluated to analyze the changes in vascular endothelial function before and after duty. The results showed that the RHI was not significantly different between before and after duty (2.12 ± 0.53 and 1.97 ± 0.50, p = 0.21) (Table 1).

The CFQ scale significantly increased after duty compared with before duty. A similar change in the VAS scores was also observed (Table 1). A significant reduction in sleep duration and HR was found after duty, while the values of rMSSD and NN50 were significantly increased (Table 1).

These results suggest that duty affects the subjective feeling of fatigue and the autonomic nervous system, but not vascular endothelial function. Since some participants felt fatigued even before duty for unknown reasons, we focused on the relationship between the degree of fatigue and vascular endothelial function.

To investigate whether fatigue or sleep deprivation may affect vascular endothelial function, we performed further analyses. Participants were divided into three groups according to RHI: normal, critical value, and abnormal group. Although ANOVA and posthoc analysis indicated no significant difference in sleep duration or CFQ scale among them, the VAS was significantly higher in the abnormal RHI group than in the normal RHI group (59 ± 13% and 46 ± 21%, respectively, p < 0.05) (Table 2). This suggests that a high degree of fatigue monitored by the VAS scale, but not the CFQ scale, may be associated with vascular endothelial dysfunction.

Compared with the normal RHI group, SBP, SDNN, and rMSSD tended to increase in the abnormal group, and the LF of the critical value group tended to increase, whereas the HR of the abnormal group tended to decrease (Table 2). Thus, we performed a binary logistic regression analysis for low RHI to adjust for SBP, SDNN, and rMSSD (Table 3). The results showed that the VAS score was independently associated with RHI in model 1 (OR 1.04 [95%CI 1.00–1.07]; p = 0.026) and remained independently associated with RHI when SBP was adjusted in model 2 (OR 1.03 [95%CI 1.00–1.07]; p = 0.044); however, no statistical significance was observed when SBP and HRV were adjusted in model 3 (OR 1.03 [95%CI 1.00–1.07]; p = 0.056) (Table 3).

In a simple regression analysis, an inverse association between sleep duration, CFQ scale (r = − 0.761, p < 0.05), and VAS (r = − 0.579, p < 0.05) were observed (Fig. 1), suggesting that the degree of fatigue is relatively parallel with lack of sleep. As for endothelial function, no significant correlation was found between the RHI and sleep duration (r = 0.045, p = 0.71) (Fig. 2a) or the CFQ scale (r = 0.002, p = 0.89) (Fig. 2b). In contrast, the RHI was significantly correlated with the VAS score (r = − 0.287, p < 0.05) (Fig. 2c). Furthermore, no association was found between AI@75 bpm and sleep or fatigue (Additional file 1: Supplementary Figure S1). In a multiple linear regression analysis, although the RHI was significantly correlated with the VAS even after adjusting for SBP, no significant association was found when SDNN, rMSSD, and HR were added for the adjustment (Table 4). RHI was negatively correlated with SBP and positively correlated with HR. These results are consistent with those from binary logistic regression analyses.