Non-restorative Sleep Caused by Autonomic and Electroencephalography Parameter Dysfunction Leads to Subjective Fatigue at Wake Time in Shift Workers

Figures

• 

  Figure 1. Experimental and shift schedule during the survey. The EEGs, ECGs, and skin temperature (Skin temp) were recorded on the night between consecutive dayshift (between day 2 and day 3). Fatigue and other subjective parameters were measured at bedtime on day 2 and at wake time on day 3, respectively. O, Day Off; D, Day shift; N, Night shift; PSQI, Pittsburgh Sleep Quality Index; MEQ, Morningness-Eveningness Questionnaire; STAI-S, state anxiety questionnaires of State-Trait Anxiety Inventory; VAS, visual analog scale; SMSH, Saint Mary's Hospital Sleep Questionnaire; KSS-J, Karolinska Sleepiness Scale (Japanese version); BP, blood pressure.

• 

  Figure 2. Comparison of subjective feeling of fatigue between the low fatigue and high fatigue groups. LFG, low fatigue group; HFG, high fatigue group. *p < 0.05, ****p < 0.0001.

• 

  Figure 3. Changes in the electroencephalography powers during sleep. (A) Changes in the natural-log transformed beta power (Ln Beta). (B) Changes in the natural-log transformed sigma power (Ln Sigma). (C) Changes in the natural-log transformed alpha power (Ln Alpha). (D) Changes in the natural-log transformed theta power (Ln Theta). (E) Changes in the natural-log transformed delta power (Ln Delta). The inset indicates the difference in the mean natural-log transformed delta power from 30 to 65 min (red-colored shaded area) after sleep onset (Ln Delta_30–65 min) between the high fatigue group (HFG) and the low fatigue group (LFG). (F) Changes in the natural-log transformed ultra-slow power (Ln US). The inset indicates the difference in the mean natural-log transformed ultra-slow power from 30 to 65 min (red-colored shaded area) after sleep onset (Ln US_30–65 min) between the HFG and LFG. *p < 0.05, **p < 0.01, ****p < 0.0001, respectively.

• 

  Figure 4. Comparison of the normalized electroencephalography powers across the three initial sleep cycles. (A) The normalized ultra-slow (US) power. (B) The normalized delta power. (C) The normalized theta power. (D) The normalized alpha power. (E) The normalized sigma power. (F) The normalized beta power. LFG, low fatigue group; HFG, high fatigue group; cycle, sleep cycle. *p < 0.05, **p < 0.01, ****p < 0.0001, respectively. #Significant difference from the LFG (p < 0.05).

• 

  Figure 5. Correlation between the deep non-rapid eye movement sleep duration in sleep cycle 1 and the electroencephalography parameters. Correlations of the stageN3 duration in sleep cycle 1 (stageN3 duration in cycle 1) with (A) the mean natural-log transformed ultra-slow powers in sleep cycle 1 (Ln US_cycle1), (B) the mean natural-log transformed ultra-slow powers from 30 to 65 min after sleep onset (Ln US_30–65 min), (C) the mean natural-log transformed delta powers in sleep cycle 1 (Ln Delta_cycle1), (D) the mean natural-log transformed delta powers from 30 to 65 min after sleep onset (Ln Delta_30–65 min), and (E) the REM latency.

• 

  Figure 6. Changes in the autonomic parameters during sleep. (A) Changes in the natural-log transformed high frequency (HF) (Ln HF). (B) Changes in the natural-log transformed low frequency (LF)/HF (Ln LF/HF). (C) Changes in the natural-log transformed LF (Ln LF). (D) Changes in the natural-log transformed very low frequency (VLF) (Ln VLF). The inset indicates the comparison of the mean Ln VLF from 30 to 65 min (red-colored shaded area) after sleep onset (Ln VLF_30–65 min) between the low fatigue group (LFG) and the high fatigue group (HFG). (E) Changes in the natural-log transformed total power (Ln TP). The inset indicates the comparison of the mean Ln TP from 0 to 30 min (red-colored shaded area) after sleep onset (Ln TP_0–30 min) between the HFG and the LFG. (F) Changes in the difference distal-proximal gradient (dDPG). The inset indicates the comparison of the mean dDPG from 30 to 250 min (red-colored shaded area) after sleep onset (dDPG_30–250 min) between the HFG and LFG. *p < 0.05, ****P < 0.0001, respectively.

• 

  Figure 7. Parameters correlated with fatigue as determined by the visual analog scale at wake time (Fatigue VAS at waketime). (A) Correlation with the stageN3 duration in sleep cycle 1 (stageN3 duration in cycle 1). (B) Correlation with the mean difference distal-proximal gradient (dDPG) from 30 to 250 min after sleep onset (dDPG_30–250 min). (C) Correlation with the mean natural-log transformed total power from 0 to 30 min after sleep onset (Ln TP_0–30 min). (D) Correlation with the rapid eye movement (REM) latency.

• 

  Figure 8. Parameters correlated with total power during the initial phase of sleep (Ln TP_0–30 min). (A) Correlation with the stageN3 duration in sleep cycle 1 (stageN3 duration in cycle 1). (B) Correlation with the mean difference distal-proximal gradient (dDPG) from 30 to 250 min after sleep onset (dDPG_30–250 min). (C) Correlation with the mean natural-log transformed ultra-slow power (Ln US) from 30 to 65 min after sleep onset (Ln US_30–65 min). (D) Correlation with the mean Ln US in sleep cycle 1 (Ln US_cycle1). (E) Correlation with the mean natural-log transformed delta power (Ln Delta) from 30 to 65 min after sleep onset (Ln Delta_30–65 min). (F) Correlation with the mean Ln Delta in sleep cycle 1 (Ln Delta_cycle1).

• 

  Figure 9. Diagram and results derived from structural equation modeling with eight observed variables. Single-headed arrows (paths) indicate the causal relationships in the model, with the variable at the tail of the arrow causing the variable at the point. Double-headed arrows indicate covariances or correlations, without a causal interpretation. Statistically, the single-headed arrows (paths) represent regression coefficients, while the double-headed arrows represent covariances. e1-8, measurement errors. Values near the arrows indicate standardized regression coefficients. *p < 0.05, **p < 0.01, #p < 0.1.

• 

  Figure 10. Modified diagram and results derived from structural equation modeling with eight observed variables. Description is the same as that for Figure 9. *p < 0.05, **p < 0.01.