Sustained attention performance during sleep deprivation associates with instability in behavior and physiologic measures at baseline

Abstract

Study objectives: To identify baseline behavioral and physiologic markers that associate with individual differences in sustained attention during sleep deprivation.

Design: In a retrospective study, ocular, electrocardiogram, and electroencephalogram (EEG) measures were compared in subjects who were characterized as resilient (n = 15) or vulnerable (n = 15) to the effects of total sleep deprivation on sustained attention.

Setting: Chronobiology and Sleep Laboratory, Duke-NUS Graduate Medical School Singapore.

Participants: Healthy volunteers aged 22-32 years from the general population.

Interventions: Subjects were kept awake for at least 26 hours under constant environmental conditions. Every 2 hours, sustained attention was assessed using a 10-minute psychomotor vigilance task (PVT).

Measurements and results: During baseline sleep and recovery sleep, EEG slow wave activity was similar in resilient versus vulnerable subjects, suggesting that individual differences in vulnerability to sleep loss were not related to differences in homeostatic sleep regulation. Rather, irrespective of time elapsed since wake, subjects who were vulnerable to sleep deprivation exhibited slower and more variable PVT response times, lower and more variable heart rate, and higher and more variable EEG spectral power in the theta frequency band (6.0-7.5 Hz).

Conclusions: Performance decrements in sustained attention during sleep deprivation associate with instability in behavioral and physiologic measures at baseline. Small individual differences in sustained attention that are present at baseline are amplified during prolonged wakefulness, thus contributing to large between-subjects differences in performance and sleepiness.

Keywords: Sleep deprivation; instability; inter-individual differences; psychomotor vigilance; variability.

Figure 1

Stratification of participants into resilient and vulnerable groups. (A) After an 8-h opportunity for sleep, subjects underwent sleep deprivation for 26 h (Protocol 1; n = 9) or 40 h (Protocol 2; n = 36) using constant routine procedures. Every 2 h, subjects completed a 10-min psychomotor vigilance task (PVT), indicated by the asterisks. PVTs scheduled during one night of sleep deprivation are shown by the box. (B) The time-course of PVT lapses in individual subjects (n = 45) revealed large inter-individual differences in sustained attention after habitual bedtime. (C) Subjects were stratified into tertiles based on each person's number of lapses during sleep deprivation. The group of participants with the fewest number of lapses was defined as resilient to sleep deprivation (n = 15; black traces), whereas the group with the greatest number of lapses was defined as vulnerable (n = 15; gray traces). In panels B-C, the vertical dotted line indicates habitual bedtime.

Figure 2

Individual differences in sustained attention during sleep deprivation are related to differences in objective sleepiness. (A) Subjects who were vulnerable to the effects of sleep deprivation on sustained attention (open circles) closed their eyes a greater percentage of the time during sleep deprivation, as compared to subjects who were resilient to sleep loss (black circles). The percentage of eyelid closure over the pupil over time (PERCLOS) was assessed during a 10-min psychomotor vigilance task (PVT) taken every 2 h. (B) During prolonged wakefulness, vulnerable subjects showed a decrease in blink rate relative to resilient subjects. (C) Despite differences in ocular measures of sleepiness, the time-course of self-rated sleepiness was similar between groups. (D) The timing of the circadian system was also similar in resilient and vulnerable subjects, as shown for the rhythm of core body temperature. In each plot, the vertical dotted line indicates habitual bedtime, which defines the boundary between baseline wakefulness and sleep deprivation. Asterisks indicate significant differences between groups by wake state (baseline versus sleep deprived). In each plot, the mean ± SEM is shown.

Figure 3

Individual differences in vulnerability to sleep deprivation are not explained by differences in homeostatic sleep regulation. (A) During baseline sleep, subjects who were resilient (black traces) or vulnerable (gray traces) to the effects of sleep deprivation on sustained attention displayed similar EEG spectral power, assessed during NREM sleep or REM sleep. (B) Similarly, during the first 8 h of recovery sleep that followed sleep deprivation, there was no difference between groups in EEG spectral power for NREM sleep or REM sleep. (C) Resilient and vulnerable groups showed an equivalent increase in spectral power in the 0-12 Hz range from baseline sleep to recovery sleep, assessed during NREM sleep in the first 2 h of the sleep episode. (D) The time-course of EEG slow wave activity (SWA) during NREM sleep was similar between resilient (black circles) and vulnerable (open circles) groups during baseline sleep and recovery sleep, suggesting that there was no difference in the buildup or dissipation of homeostatic sleep pressure. In each plot, the mean ± SEM is shown, and data are shown for the central EEG derivation.

Figure 4

Individual differences in psychomotor vigilance task (PVT) performance during baseline wakefulness and sleep deprivation. PVT performance measures are shown in vulnerable subjects (open circles) and resilient subjects (black circles) during prolonged wakefulness. During sleep deprivation, vulnerable individuals showed a much greater increase than resilient individuals in (A) lapses, (B) mean reaction time (RT), (C) standard deviation (SD) of RT, and (D) the number of consecutive RTs that differed by greater than 250 ms. In the right column, PVT measures during the daytime are re-plotted on a different scale to highlight baseline differences between resilient and vulnerable groups. In the left column, asterisks indicate significant differences between groups by wake state (baseline versus sleep-deprived). The vertical dotted line indicates habitual bedtime, which defines the boundary between baseline wakefulness and sleep deprivation. In the right column, asterisks show significant differences between groups at each time point (t-test, P < 0.05). In each plot, the mean ± SEM is shown.

Figure 5

Individual differences in heart rate and its variability associate with differences in psychomotor vigilance task (PVT) performance during sleep deprivation. (A) Irrespective of time elapsed since wake, individuals who were vulnerable to the effects of sleep deprivation (open circles) on sustained attention exhibited a lower heart rate than resilient individuals (black circles) during the PVT. (B) Vulnerable participants also displayed greater variability in heart rate than resilient individuals, as assessed by the standard deviation of sinus normal to normal inter-beat intervals (SDNN). Based on frequency domain measures of the RR-interval time series, vulnerable subjects exhibited greater spectral power in (C) very-low-frequency (VLF, ≤ 0.04 Hz), (D) low-frequency (LF, 0.04-0.15 Hz), and (E) high-frequency (HF, 0.15-0.40 Hz) bands, as compared to resilient subjects. Asterisks indicate significant differences between groups by wake state (baseline versus sleep deprived). Where a significant interaction between group and wake state was not observed, number symbols (#) indicate a significant main effect of group on heart rate variability. In each plot, the mean ± SEM is shown.

Figure 6

Individual differences in EEG spectral power associate with differences in psychomotor vigilance task (PVT) performance during prolonged wakefulness. (A) During the daytime, individuals who were vulnerable to the effects of sleep deprivation on sustained attention (gray trace) exhibited greater EEG spectral power in the high-frequency theta range (6.0-7.5 Hz) range while taking the PVT, as compared to resilient individuals (black trace). (B) After habitual bedtime, vulnerable individuals continued to show greater high-frequency theta activity than resilient individuals, as well as more spectral power in the delta band. (C) Vulnerable subjects (open circles) showed a greater increase in delta power during sleep deprivation relative to resilient subjects (black circles), whereas (D) EEG theta power was consistently higher in vulnerable participants at all times of day, regardless of time elapsed since wake. (E) Vulnerable individuals showed greater variability in delta power, as well as (F) theta power, relative to resilient individuals, assessed by the epoch-to-epoch standard deviation (SD) of spectral power within each PVT. In panels C-F, the vertical dotted line indicates habitual bedtime, which defines the boundary between baseline wakefulness and sleep deprivation. Asterisks indicate significant differences between groups by wake state (baseline versus sleep deprived). Where a significant interaction between group and wake state was not observed, number symbols (#) indicate a significant main effect of group on EEG measures. In each plot, the mean ± SEM is shown.

Figure 7

Baseline individual differences in psychomotor vigilance task (PVT) performance variability and heart rate variability associate with relative vulnerability to sleep deprivation. Subjects were stratified into top and bottom tertiles (n = 15 in each group) based on their standard deviation of PVT reaction times, or by RR-interval spectral power in the very-low-frequency band (VLF, ≤ 0.04 Hz), assessed during a single 10-min PVT taken in the middle of the habitual wake period. (A) Individuals with high variability in PVT reaction time (open circles) showed a greater increase in attentional lapses during sleep deprivation than individuals with low variability (black circles). (B) Similarly, subjects with high VLF spectral power (open circles) during the daytime showed a greater decline in PVT performance during prolonged wakefulness than subjects with low VLF spectral power (black circles). (C) Individuals with high variability in daytime PVT performance showed a greater increase in eye closures during sleep deprivation, (D) as did individuals with high spectral power in the VLF band, relative to individuals with low variability in these measures. In each panel, the vertical dotted line indicates habitual bedtime, which defines the boundary between baseline wakefulness and sleep deprivation. Asterisks indicate significant differences between groups by wake state (baseline versus sleep deprived). In each plot, the mean ± SEM is shown.

Figure 8

Baseline individual differences in psychomotor vigilance task (PVT) performance variability and heart rate variability are reproducible across study visits. A subset of individuals that was characterized as resilient (n = 12) or vulnerable (n = 13) to the effects of sleep deprivation on sustained attention participated in a follow-up study at least 5 months after their initial visit to the laboratory. Participants were required to keep a fixed sleep schedule with 8 h time-in-bed per night prior to the first laboratory visit, but they were free to choose their sleep and wake times prior to the second visit. We compared PVT performance and heart rate variability during the middle of the habitual wake period between study visits. Regardless of prior sleep history (fixed versus ad libitum sleep), vulnerable individuals (white bars) exhibited more variable response times than resilient individuals (black bars), as assessed by (A) the standard deviation of PVT reaction time (RT), and (B) the number of consecutive RTs that differed by greater than 250 ms. (C) During both study visits, resilient subjects displayed a higher heart rate than vulnerable subjects, and both groups showed an increase in heart rate during the second visit. (D) Vulnerable subjects exhibited greater RR-interval VLF spectral power than resilient subjects, irrespective of differences in sleep history. Number symbols (#) show main effects of group on PVT and heart rate measures. In each plot, the mean ± SEM is shown.