SEARCH

SEARCH BY CITATION

Keywords:

  • chronic sleep restriction;
  • modeling;
  • partial sleep deprivation;
  • performance;
  • recovery;
  • sleep deprivation;
  • sleep restriction;
  • sleep

SUMMARY

  1. Top of page
  2. SUMMARY
  3. References

Daytime performance changes were examined during chronic sleep restriction or augmentation and following subsequent recovery sleep. Sixty-six normal volunteers spent either 3 (n = 18), 5 (n= 16), 7 (n = 16), or 9 h (n = 16) daily time in bed (TIB) for 7 days (restriction/augmentation) followed by 3 days with 8 h daily TIB (recovery). In the 3-h group, speed (mean and fastest 10% of responses) on the psychomotor vigilance task (PVT) declined, and PVT lapses (reaction times greater than 500 ms) increased steadily across the 7 days of sleep restriction. In the 7- and 5-h groups speed initially declined, then appeared to stabilize at a reduced level; lapses were increased only in the 5-h group. In the 9-h group, speed and lapses remained at baseline levels. During recovery, PVT speed in the 7- and 5-h groups (and lapses in the 5-h group) remained at the stable, but reduced levels seen during the last days of the experimental phase, with no evidence of recovery. Speed and lapses in the 3-h group recovered rapidly following the first night of recovery sleep; however, recovery was incomplete with speed and lapses stabilizing at a level comparable with the 7- and 5-h groups. Performance in the 9-h group remained at baseline levels during the recovery phase. These results suggest that the brain adapts to chronic sleep restriction. In mild to moderate sleep restriction this adaptation is sufficient to stabilize performance, although at a reduced level. These adaptive changes are hypothesized to restrict brain operational capacity and to persist for several days after normal sleep duration is restored, delaying recovery.


Introduction

  1. Top of page
  2. SUMMARY
  3. References

Chronic sleep restriction is endemic in modern society. Americans are sleeping less (Bliwise et al. 1992; Webb and Agnew 1975) and, in absolute terms, a large fraction is reporting daily sleep substantially below the recommended optimum of 8 h per night (National Sleep Foundation 2002; Sleep in America Poll), although the true optimum may be higher (Coren 1997; Palinkas et al. 1995) or lower (Harrison and Horne 1995; Kripke et al. 2002). Sleep deprivation and chronic sleep restriction degrade health, safety, productivity (individual and societal) and quality of life (Bonnet and Arand 1995; Leger 1994; Mitler et al. 1988; Pilcher and Huffcutt 1996; Spiegel et al. 1999). In contrast to the apparent endemic nature of chronic sleep restriction, acute total sleep deprivation (TSD) occurs infrequently outside of the sleep laboratory.

Despite the prevalence of sleep restriction in modern society, few studies have evaluated the performance effects of chronic sleep restriction (e.g. Carskadon and Dement 1981; Dinges et al. 1997, 1999; Friedman et al. 1977; Horne and Wilkinson 1985; Kuo et al. 1998; Mullaney et al. 1977; Webb and Agnew 1974). Most studies to determine the behavioral effects of sleep loss have used acute TSD as the experimental manipulation. TSD studies are more efficient than chronic sleep restriction studies as the former produce larger behavioral decrements over a shorter period of time. Because of the paucity of well-controlled sleep restriction studies, findings from acute TSD studies, and specifically findings with respect to the accumulation and discharge of pressure for electroencephalographic (EEG) slow wave activity, have generally been used in the development of mathematical models to predict performance on the basis of prior sleep–wake history (Achermann et al. 1993; Akerstedt and Folkard 1995, 1996, 1997; Borbely 1982; Borbely et al. 1989; Folkard and Akerstedt 1991; Jewett and Kronauer 1999; Jewett et al. 1999). These models, both in terms of degradation and recovery functions, are anchored at the extremes of normal sleep on the one hand and TSD on the other with little objective data on what goes on in between (i.e. during restricted sleep). In sum, little is known about the effect of chronic sleep restriction on objectively measured alertness and performance and on the time course of subsequent recovery.

The purpose of the present study was to empirically determine the effects of several levels of restricted and one level of augmented sleep over seven consecutive days on objective and subjective alertness and objective performance, and to determine the extent to which 3 days of subsequent recovery sleep restored performance and alertness to baseline levels – a sleep dose–response study.

The data reported here are a subset of data collected in a larger study analysed and published as a US Department of Transportation Report (Balkin et al. 2000). In the present paper we report findings for psychomotor vigilance task performance (PVT), sleep latency, and subjective sleepiness. Of the measures taken in the larger study, the PVT was chosen for this paper because it was the most sensitive to the effects of sleep restriction and was the least subject to learning effects (Balkin et al. 2000).

Subjects

  1. Top of page
  2. SUMMARY
  3. References

Sixty-six volunteers (16 women, age 24–55, mean = 43 years; and 50 men, age 24–62, mean = 37 years) participated. All subjects held valid Commercial Motor Vehicle (CMV) drivers' licenses. Subjects were in good general health as determined by medical history and medical examination and were free of neurological diseases, psychiatric disorders, sleep disorders, and drug or alcohol addiction. They did not use nicotine in any form and reported consuming no more than 300–400 mg caffeine per day. Subjects were medication-free (including over-the-counter medications) beginning 48 h prior to the study, with the exception that female subjects could continue birth control medications.

Design

  1. Top of page
  2. SUMMARY
  3. References

Volunteers spent 14 days in-residence in the laboratory (Fig. 1); a ‘day’ is defined as the time from the beginning of one scheduled sleep period to the beginning of the next. The first 3 days (T1, T2 and B) were adaptation and training (T1 and T2) and baseline (B) and subjects were required to be in bed from 23:00 to 07:00 h [8 h required time in bed (TIB)]. On the third day (B), baseline measures were taken. Beginning on the fourth day and continuing for a total of 7 days (E1–E7) subjects were in one of four sleep conditions [9 h required TIB (22:00–07:00 h), 7 h required TIB (24:00–07:00 h), 5 h required TIB (02:00–07:00 h), or 3 h required TIB (04:00–07:00 h)], effectively one sleep augmentation condition, and three sleep restriction conditions. The purpose of these conditions was to produce differing levels of sleep restriction/augmentation relative to the adaptation and training and baseline days. Beginning on the eleventh day and continuing for a total of 3 days (R1–R3) subjects were again required to be in bed from 23:00 to 07:00 h (8 h required TIB). These last 3 days constituted recovery. Subjects obtained a final night of 8 h TIB prior to release from the study; however, no testing occurred following this fourth night of recovery sleep. Throughout all phases of the study, lights on for all subjects was 07:00 h. Subjects were not permitted any other time in bed or opportunity for sleep except as required by the periodic sleep latency tests (described below).

image

Figure 1. Study experimental design, showing nightly time in bed across days (adaptation/training, baseline, experimental phase, recovery phase).

Download figure to PowerPoint

Test instruments and measures

  1. Top of page
  2. SUMMARY
  3. References

Psychomotor vigilance test

  1. Top of page
  2. SUMMARY
  3. References

The PVT measures simple reaction time to a visual stimulus, presented approximately 10 times/minute (interstimulus interval varied from 2 to 10 s in 2-s increments) for 10 min and implemented in a thumb-operated, hand-held device (Dinges and Powell 1985). Subjects attended to the LED timer display on the device and pressed the response button with the preferred thumb as quickly as possible after the appearance of the visual stimulus. The visual stimulus was the LED timer turning on and incrementing from 0 at 1-ms intervals. In response to the subject's button press, the LED timer display stopped incrementing and displayed the subject's response latency for 0.5 s, providing trial-by-trial performance feedback. At the end of this 0.5-s interval the display turned off for the remainder of the foreperiod preceding the next stimulus. Foreperiods varied randomly from 2 to 10 s. Dependent measures, averaged or summed across the 10-min PVT session, included mean speed (reciprocal of average response latency), number of lapses (lapse = response latency exceeding 500 ms), and mean speed for the fastest 10% of all responses.

Polysomnography

  1. Top of page
  2. SUMMARY
  3. References

Polysomnographic (PSG) measures [EEG (C3 and C4); EOG (outer canthi of each eye), EMG (mental/submental)], and EKG (from just below left and right clavicle) were recorded continuously throughout the study using Medilog 9000-II magnetic cassette recorders (Oxford Instruments, Largo, FL, USA). Raw data were digitized, and both night-time sleep and sleep latency tests (described below) were scored in accordance with Rechtschaffen and Kales (1968) criteria using Eclipse software (Stellate Systems, Westmont, Quebec, Canada).

Night-time sleep

  1. Top of page
  2. SUMMARY
  3. References

Six technicians, whose inter-rater reliabilities were at least 85% compared with the scoring of a diplomate of the American Board of Sleep Medicine (TJB), scored night-time sleep periods (defined as lights out to lights on). Dependent measures included minutes of individual sleep stages [1, 2, slow wave sleep (SWS) and REM] and minutes of total sleep time (TST) (sum of minutes spent in all sleep stages).

Sleep latency test

  1. Top of page
  2. SUMMARY
  3. References

Subjects were placed in bed in a quiet, darkened room and instructed to close their eyes and not resist the urge to fall asleep. Staff monitored PSG signals from outside of the bedroom using Oxford Mentor systems. To decrease the likelihood of premature test termination during ambiguous stage 1, the sleep latency test (SLT) was terminated immediately after the onset of stage 2 sleep (or after 20 min without sleep onset) by a staff member opening the bedroom door, turning on the lights, and announcing that the test was over. As stage 1 sleep does not appear to confer recuperative benefit in otherwise normal, healthy adults (Wesensten et al. 1999), the potential accumulation of up to 40 min stage 1 sleep daily (20 min per SLT × 2 SLTs per day) was not expected to affect performance. SLT schedules were staggered by 25 min for subject roommates, so that each could be tested in the bedrooms individually. For purposes of analyses, sleep latency was re-scored off-line from lights out to the first 30 s of stage 1 sleep.

Subjective alertness/sleepiness

  1. Top of page
  2. SUMMARY
  3. References

The Stanford Sleepiness Scale (SSS; Hoddes et al. 1973) assessed subjective sleepiness on a single-item scale ranging from 1 (‘feeling active and vital; alert; wide awake’) to 7 (‘almost in reverie; sleep onset soon; losing struggle to remain awake’). The dependent measure was the subject's sleepiness rating.

Procedure

  1. Top of page
  2. SUMMARY
  3. References

Subjects reported to the Division of Neuropsychiatry, Walter Reed Army Institute of Research at 10:00 h on the day prior to T1. After being provided with verbal and written descriptions of study procedures and rules, subjects were individually informed of the sleep schedule to which their group (of two to four subjects) was being assigned and electrodes for ambulatory PSG (Oxford Medilog 9200-II), including EOG, EMG, C3, C4, O1, O2, and EKG were applied. Subjects then underwent training on the various performance tasks. At 18:00 h, they were transported to the Johns Hopkins Bayview General Clinical Research Center (GCRC, Baltimore, MD, USA) where they resided until the end of the study. Throughout the study, meals were served at 08:30, 12:30, and 17:30 h, with snacks and beverages available ad libitum between performance tests. Vital signs (blood pressure, pulse, and tympanic temperature) were recorded periodically for purposes of checking general health status. Subjects did not use/consume nicotine or caffeine-containing products during the study; random urine drug screens verified compliance. Use of medications during the study (e.g. acetaminophen for headache) was allowed at the discretion of the attending physician. For all women enrolled in the study, serum pregnancy tests performed at the beginning of the study were negative.

Same-sex subject pairs were assigned to share 2-person hospital-style bedrooms. T1 and T2 were devoted to training on the performance tests and familiarization with study procedures. Baseline testing commenced on the morning of the third day (B) and testing continued for the duration of the study (E1–E7, R1–R3). On the morning of R4 electrodes were removed shortly after awakening, and subjects were debriefed and released from the study. No testing occurred on R4.

Test schedule

  1. Top of page
  2. SUMMARY
  3. References

The PVT and SSS were administered together as a battery four times per day (09:00, 12:00, 15:00, and 21:00 h); the battery included other tests not reported here (see Balkin et al. 2000). The sleep latency test was administered at 09:40 and 15:30 h for all groups.

Subjects in the 3- and 5-h TIB groups performed an additional battery at 00:00 h and 02:00 h to occupy their additional time awake. The PVT and SSS were administered in this battery; however, as data from the 00:00 and 02:00 h sessions were not common to all TIB groups, these data were not included in the statistical analyses reported below.

Statistical analyses

  1. Top of page
  2. SUMMARY
  3. References

The PVT, SLT, and SSS data were analysed using a mixed analysis of variance (anova) for TIB group (3-, 5-, 7-, or 9-h), day (B, E1, E2, E3, E4, E5, E6, E7, R1, R2, and R3), and time of day [09:00, 12:00, 15:00, and 21:00 h for PVT and SSS; 09:30 and 15:40 h for multiple sleep latency test (MSLT)] with repeated measures on the latter two factors. The time of day factor was included to help determine whether sleep group-specific changes across days were mediated by phase shifts in the circadian rhythm of performance, and to help determine whether sleep restriction effects were primarily manifested at particular times of day. These possibilities would be suggested by a three-way Sleep Group × Day × Time of Day interaction. Therefore, in results reported below, only significant three-way interactions are reported (two-way interactions and main effects for time of day are not reported). Night-time sleep data were analysed using a two-way mixed anova for TIB group and day, with repeated measures on the day factor. Procedures outlined by Kirk (1995) were followed for interpretation of interactions – first, simple main effects analyses were applied to all significant Sleep Group–Day interactions (i.e. simple main effects of day separately for each sleep group, and simple main effects of sleep group separately for each day). Next, post hoc comparisons (Tukey's Honestly Significant Difference procedure or ‘HSD’–Kirk 1995) were conducted for each significant simple main effect to specify differences among mean values. Greenhouse–Geisser corrections (Kirk 1995) were applied to all repeated measures effects. Unless otherwise noted, a significance level of P ≤ 0.05 was used for all statistical analyses.

Results

  1. Top of page
  2. SUMMARY
  3. References

Due to technical difficulties, one or more sessions of PVT performance data were lost for some subjects from each sleep group. The final n for each sleep group and demographic data are reported in Table 1. Neither the distribution of females and males (chi-square analyses) nor the mean age (one-way analyses of variance) differed among TIB groups (both analyses P > 0.05).

Table 1.  Demographic data for each time in bed (TIB) group
TIB groupNo. of females/mean age (year)No. of males/mean age (year)
3 h (n = 13)5/39.28/39.6
5 h (n = 13)2/45.511/32.5
7 h (n = 14)4/43.810/38.6
9 h (n = 16)4/47.012/37.4

Total sleep time

  1. Top of page
  2. SUMMARY
  3. References

Figure 2 shows mean TST (TST = sum of stages 1, 2, SWS and REM) for each group across B, E1–E7, and R1–R3. TST increased significantly in the 9-h group and decreased significantly in the 3-, 5- and 7-h groups across the sleep restriction/augmentation phase (E1–E7) compared with baseline (B) (group, F3,61 = 486.53, P = 0.0000; night, F10,610 = 281.10, P = 0.0000; Group × Night, F(30, 610) = 141.83, P = 0.0000). Average TST over the 7 days of sleep restriction/augmentation were 7.93 h for the 9-h TIB group, 6.28 h for the 7-h TIB group, 4.66 h for the 5-h TIB group, and 2.87 h for the 3-h TIB group.

image

Figure 2. Mean total daily sleep durations in minutes (and hours – right-hand ordinate) as a function of time in bed group. Error bars (standard error of the mean) are included but are obscured by data points.

Download figure to PowerPoint

Table 2 lists F-values and corresponding Tukey HSD comparisons for group and day simple effects. TST significantly differed among all TIB groups on nights E1–E7 but not on B or R1–R3. Within each TIB group, nightly sleep amounts did not change across E1–E7. For all groups, TST amounts appeared to return to baseline values during recovery, i.e. B was not different from R1–R3 for any group.

Table 2.  Analyses of variance results of simple effects (TIB group, day) for total sleep time (TST). (a) Simple effects of TIB group for each day, (b) Simple effects of day for each TIB group
(a)
DayFDFnumDFdenomPHSD value
  • TIB, time in bed; TST, total sleep time, HSD, honestly significant difference.

  • *

    HSD values are not applicable to non-significant simple effects.

B1.073610.3675*
E1826.443610.000018.32
E22239.873610.000011.35
E3523.543610.000022.50
E4691.373610.000020.05
E5537.503610.000022.00
E6829.633610.000017.87
E7434.823610.000024.07
R10.843610.4766*
R20.553610.6506*
R32.223610.0944*
(b)
TIB groupFDFnumDFdenomPHSD value
3 h566.72106100.000024.30
5 h143.19106100.000025.77
7 h12.60106100.000026.62
9 h20.80106100.000025.77

Sleep stages

  1. Top of page
  2. SUMMARY
  3. References

Amounts of stage 1, 2 and REM sleep changed in a dose-dependent fashion during the experimental phase (E1–E7); stages 1, 2, and REM significantly decreased in the 3- and 5-h TIB groups while stages REM and 1 significantly increased in the 9-h TIB group (group simple effects, P < 0.05). Amounts of stages 1 and 2 also decreased in the 7-h TIB group, albeit non-significantly (P > 0.05). Sleep stage amounts returned to baseline values during recovery for all groups (B not different from R1–R3; Tukey HSD, P > 0.05). There were no group differences on B or R1–R3 (group simple effects, P > 0.05). The SWS amounts failed to vary significantly across experimental days as a function of TIB group (P > 0.05).

Psychomotor vigilance test

  1. Top of page
  2. SUMMARY
  3. References

Mean speed

  1. Top of page
  2. SUMMARY
  3. References

Figure 3 shows mean PVT response speed {[1/mean reaction time (RT)] × 1000} as a function of TIB group and day (collapsed across time of day), and Table 3 lists F-values and corresponding Tukey HSD comparisons for TIB group and day simple effects. Table 4 lists standard deviations for mean speed as a function of day and TIB group. Response speed decreased across experimental days in a TIB dose-dependent manner for the 3-, 5-, and 7-h TIB groups (group F3,52 = 5.72, P = 0.0018; day F10,520 = 19.01, P = 0.0000); Group × Day F30,520 = 6.64, P = 0.0000; simple effects for day at each group are given in Table 3. Speed for the 9-h TIB group did not vary across days. For the 3-h TIB group, speed on days E2 through E7 was lower than B; within the experimental phase, E3–E7 were lower than E1 and E2; E6 was lower than E1–E3; and E7 was lower than E1–E5. Speed increased significantly in the 3-h TIB group from E7 to R1, but no further speed increases were evident during recovery (R1 not different from R2 and R3); and speed failed to recover to baseline levels in the 3-h TIB group (R1–R3 < B). For the 5-h TIB group, speed was significantly impaired on E3–E7 compared with B; within the experimental phase E5–E7 were significantly lower than E1 and E2. Although speed appeared to increase from E7 to recovery days, the difference was not significant (E7 not different from R1–R3). Speed in the 5-h TIB group failed to recover (R1–R3 < B). For the 7-h group, speed was lower on R2 and R3 compared with B and E1. Within days, during the experimental phase speed for the 3-h TIB group was significantly lower than speed for 9-h group on E2–E7, significantly lower than speed in the 7-h group on E3–E7, and significantly lower than speed in the 5-h group on E7. Speed for the 5-h group was significantly lower than that for the 9-h group on E5. Speed did not differ between the 7- and 9-h TIB groups during the experimental phase. During the recovery phase, speed for the 3-h TIB group was significantly lower than speed for the 9-h TIB group on R1 and R3. Speed did not differ among the 5-, 7-, and 9-h TIB groups on any recovery day.

image

Figure 3. Mean psychomotor vigilance task speed (and standard error) across days as a function of time in bed group.

Download figure to PowerPoint

Table 3.  Analyses of variance results of simple effects (TIB group, day) for PVT mean speed. (a) Simple effects of TIB group for each day, (b) Simple effects of day for each TIB group
(a)
DayFDFnumDFdenomPHSD value
  • TIB, time in bed; TST, total sleep time; HSD, honestly significant difference; PVT, psychomotor vigilance task.

  • *

    HSD values are not applicable to non-significant simple effects.

B0.183520.9079*
E11.873520.1457*
E22.953520.04120.5909
E36.053520.00130.7127
E46.953520.00050.7634
E58.893520.00010.7417
E69.773520.00000.7633
E714.203520.00000.7524
R13.743520.01650.7216
R22.623520.06020.7790
R33.153520.03250.8091
(b)
TIB groupFDFnumDFdenomPHSD value
3 h26.11105200.00000.3824
5 h7.14105200.00010.3824
7 h3.25105200.01770.3685
9 h0.11105200.9669*
Table 4.  Standard deviations of mean speed for each day and time in bed (TIB) group
DayTIB Group
3-HR5-HR7-HR9-HR
B0.50320.71730.57110.528
E10.69720.86670.53350.5001
E20.74120.81440.52750.4632
E30.94710.89250.62750.5028
E41.06480.98350.72130.4626
E50.95431.10220.64690.4431
E60.92051.07150.77280.4119
E71.01311.01490.76190.4723
R10.88250.95430.75750.4673
R20.72271.01930.87990.5169
R30.82961.11190.85440.4492

The three-way Group × Day × Time of Day interaction was not significant (P = 0.1018).

Number of lapses

  1. Top of page
  2. SUMMARY
  3. References

Figure 4 shows mean number of lapses across days as a function of TIB group, and Table 5 lists F-values and corresponding Tukey HSD comparisons for TIB group and day simple effects. Lapses increased across the experimental phase in the 3- and 5-h groups but not in the 7- and 9-h groups (group F3,52 = 7.29, P = 0.0004; day F10,520 = 15.84, P = 0.0000; Group × Day F30,520 = 6.90, P = 0.0000). For the 3-h group, more lapses occurred on E3–E7 compared with B; on E3–E7 compared with E1 and E2; on E5–E7 compared with E3; and on E7 compared with E3–E5. Lapses decreased from E7 to recovery (E7 > R1–R3) but failed to return to baseline levels (R1 and R3 > B). For the 5-h TIB group, more lapses occurred on E6 and E7 compared with B; lapses failed to recover (R2 and R3 > B). During the experimental phase more lapses occurred in the 3-h TIB group than (a) the 9-h group on E3–E7; (b) the 7-h group on E4–E7; and (c) the 5-h group on E6 and E7. Mean number of lapses did not differ among the 5-, 7- and 9-h TIB groups during the experimental phase. During the recovery phase, more lapses occurred in the 3-h TIB group compared with the 9- h group on R1. Lapses did not differ among the 5-, 7-, and 9-h TIB groups on any recovery day.

image

Figure 4. Mean number of lapses on the psychomotor vigilance task (and standard error) across days as a function of time in bed group.

Download figure to PowerPoint

Table 5.  Analyses of variance results of simple effects (TIB group, day) for PVT number of lapses. (a) Simple effects of TIB group for each day, (b) Simple effects of day for each TIB group
(a)
DayFDFnumDFdenomPHSD value
  • TIB, time in bed; HSD, honestly significant difference; PVT, psychomotor vigilance task.

  • *

    HSD values are not applicable to non-significant simple effects.

B0.383520.7660*
E11.633520.1946*
E22.933520.04214.3497
E36.553520.00085.9189
E47.753520.00026.3229
E511.223520.00006.0335
E610.083520.00007.2513
E716.713520.00006.6998
R13.583520.01985.6057
R22.423520.0767*
R33.093520.03506.7452
(b)
TIB groupFDFnumDFdenomPHSD value
3 h29.69105200.00003.9439
5 h3.69105200.00513.9439
7 h1.21105200.3082*
9 h0.05105200.9972*

A significant three-way Group × Day × Time of Day interaction (F(30,520) = 1.67, P = 0.0235) indicated that the time of day effect (more lapses at 09:00 h vs. 21:00 h) was greatest in the 3-h group, and that these time of day differences in the 3-h group increased across the experimental phase. There was no evidence that the timing of the peak or trough in number of lapses shifted across experimental days for any TIB group.

Mean speed – fastest 10% of responses

  1. Top of page
  2. SUMMARY
  3. References

To verify that sleep dose-dependent effects on mean speed were not due solely to an increased number of lapses, mean speed for the fastest 10% of responses also was analysed. Figure 5 shows mean speed for the fastest 10% of response as a function of TIB group and day (collapsed across time of day), and Table 6 lists results of the anova. The pattern for the fastest 10% of responses was similar to that found for mean speed: response speed decreased across experimental days in a TIB dose-dependent manner for the 3- and 5-h TIB groups but not for the 7- or 9-h groups. For the 3-h TIB group, speed on days E3–E7 was lower than on B; for the 5-h TIB group, speed on days E2–E7 was lower than on B. For both the 3- and 5-h TIB groups, speed increased non-significantly from E7 to R1; speed among days R1–R3 did not differ, and speed on all recovery days was lower than baseline (R1–R3 < B). During the experimental phase, speed for the 3-h TIB group was significantly lower than speed for the 7 and 9-h TIB groups on E3–E7; but mean speed did not differ among the 5-, 7-, and 9-h TIB groups during the experimental phase. During the recovery phase, although mean speed for the 3-h TIB group appeared to remain lower than that of the 9-h TIB group, these differences were not significant. Mean speed did not differ among the 5-, 7-, and 9-h TIB groups on any recovery day.

image

Figure 5. Mean psychomotor vigilance task speed for the fastest 10% of responses (and standard error) across days as a function of TIB group.

Download figure to PowerPoint

Table 6.  Analyses of variance results for PVT mean fastest 10% of responses. (a) Main effects and interactions, (b) Simple effects of TIB group at each day, (c) Simple effects of day for each TIB group
(a)
EffectFDFnumDFdenomP 
  • TIB, time in bed; HSD, honestly significant difference, PVT, psychomotor vigilance task.

  • *

    HSD values are not applicable to non-significant simple effects.

TIB Group (G)4.57 3  520.0065 
Day (D)11.8110 5200.0000 
Time of day (T)9.48 3 1560.0001 
G × D2.7530 5200.0004 
G × T0.55 9 1560.7799 
D × T0.983015600.4767 
G × D × T0.879015600.7077 
(b)
DayFDFnumDFdenomPHSD value
B1.203520.3185*
E12.123520.1089*
E22.323520.0857*
E35.033520.00390.4828
E46.253520.00110.4942
E55.863520.00160.4439
E66.983520.00050.4702
E79.473520.00000.4399
R12.613520.06090.4668
R22.913520.04310.4818
R33.013520.03830.5291
(c)
TIB groupFDFnumDFdenomPHSD value
3 h9.93105200.00000.2546
5 h6.72105200.00000.2546
7 h1.88105200.0924*
9 h0.33105200.9070*

Objective and subjective sleepiness

  1. Top of page
  2. SUMMARY
  3. References

Figure 6 shows latency to the first 30 s of stage 1 sleep from the SLT, and Fig. 7 shows sleepiness ratings from the SSS across days as a function of TIB group (means collapsed across time of day).

image

Figure 6. Mean latency to the first 30 s of stage 1 sleep (and standard error) across days as a function of time in bed group.

Download figure to PowerPoint

image

Figure 7. Mean Stanford Sleepiness Scale scores (and standard error) across days as a function of time in bed group.

Download figure to PowerPoint

Daytime sleep latency

  1. Top of page
  2. SUMMARY
  3. References

On the baseline day (B), average sleep latency was 6.3 min ± 0.64 SEM across all groups. Mean sleep latency did not differ between TIB groups on Baseline (TIB group simple effect, P > 0.05). Latency decreased across experimental days for the 3- and 5-h TIB groups, but not for the 7- and 9-h TIB groups (group F3,50 = 7.61, P = 0.0003; day F(10,500) = 4.29, P = 0.0002; Group × Day: F(300,500) = 2.31, P = 0.0014; day simple effects: 3-h, F(10, 500) = 4.29, P = 0.0002; 5-h, F(10,500) = 4.29, P = 0.0002; P > 0.05 for 7- and 9-h TIB groups). For the 3-h TIB group, latency on E5 was significantly shorter than B (Tukey HSD = 3.65, P < 0.05). Although latency appeared to increase from the experimental to the recovery phase in the 3-h group, differences between experimental and recovery days were not significant (Tukey HSD, P > 0.05); also, although it appeared that sleep latencies failed to recover to baseline levels, R1–R3 were not significantly different from B (Tukey HSD, P > 0.05). Significant time (F1,50 = 23.03, P = 0.0000) and Group × Time (F3,50 = 9.3, P = 0.0000) effects indicated dose-ordered differences in sleep latency (9- > 7- > 5- > 3-h) in the morning but no sleep latency differences in the afternoon.

Stanford sleepiness scale

  1. Top of page
  2. SUMMARY
  3. References

Sleepiness ratings increased significantly during the experimental phase in the 3-h group but did not change in the 5-, 7-, and 9-h groups (group F3,56 = 3.05, P = 0.0359; day F(10,560) = 6.58, P = 0.0000; Group × Day: F(30, 560) = 4.42, P = 0.0000; day simple effect for 3-h, F(10,560) = 18.28, P = 0.0000; day simple effects for 5-, 7-, and 9-h groups, P > 0.05). For the 3-h group, sleepiness ratings on E2–E7 were significantly higher than B, and E5–E6 were higher than E1 (Tukey HSD = 0.44, P < 0.05). Sleepiness ratings decreased significantly from experimental to recovery days (R1–R3 < E2–E7; Tukey HSD P < 0.05) and returned to baseline during the latter (R1–R3 not different from B; Tukey HSD P > 0.05). Within days, sleepiness ratings for the 3-h group were significantly higher than other TIB groups as follows: E2 (F3,56 = 4.54, P = 0.0064), 3 > 9; E3 (F3,56 = 4.28, P = 0.0086), 3 > 7, 9; E4 (F3,56 = 2.79, P = 0.0488), Tukey HSD, NS; E5 (F3,56 = 6.01, P = 0.0013) and E6 (F3,56 = 5.46, P = 0.0023), 3 > 7, 9; and E7 (F3,56 = 3.97, P = 0.0123), 3 > 7. No differences among TIB groups were found on B, E1, and R1–R3 (group simple effects, P > 0.05).

Significant effects for time of day (F3,168 = 4.27, P = 0.0121), and Group × Time of Day (F9,168 = 2.39, P = 0.0255) indicated that dose-dependent differences in sleepiness scores were maximal in the morning.

Discussion

  1. Top of page
  2. SUMMARY
  3. References

Seven days of sleep restriction degraded psychomotor vigilance performance in a sleep-dose-dependent manner. With mild to moderate sleep restriction (7- and 5-h TIB), performance initially declined and, after a few days, appeared to stabilize at a lower-than-baseline level for the remainder of the sleep restriction period. In contrast, with severe sleep restriction (3-h TIB) performance declined continuously across the sleep restriction period, with no apparent stabilization of performance. Sleep augmentation (9-h TIB) had no effect on performance over the 7-day experimental period. Based on these findings, it appears that the inflection point (i.e. the minimum amount of nightly sleep required to achieve a state of equilibrium in which daytime alertness and performance can be maintained at a stable, albeit reduced, level) is approximately 4 h per night. If less than 4 h of sleep per night is obtained, daily decrements in performance capacity may be unavoidable – at least across a 7-day period of sleep restriction.

Core vs. optional sleep

  1. Top of page
  2. SUMMARY
  3. References

These findings are reminiscent of Horne's hypothesis that a minimum amount of nightly sleep (approximately 4.5 h per night) is required to satisfy the brain's physiological need for recuperation (core sleep), and that any sleep obtained in excess of this minimum is essentially appetitive (Horne 1988). According to this view, utilization of spare cerebral capacity enables the brain to perform well with less than 8 h of sleep each night. The possibility of such spare cerebral capacity is consistent with the suggestion of Drummond et al. (2000) that cognitive performance during sleep deprivation involves recruitment of resources from additional, non-task specific – and therefore relatively non-fatigued – brain regions.

Although the present findings tend to support Horne's hypothesis that there is a physiological limit to the ability to tolerate sleep restriction and that this limit is approximately 4 h of sleep per night, these findings also suggest a more continuous gradient of recuperative value than that suggested by the core sleep and optional sleep designations. For example, our finding that PVT performance was decremented during recovery following 7 days of a mild, 1-h reduction of nightly TIB – from 8 to 7 h – suggests that meaningful recuperation continues to accrue as sleep durations are extended beyond the hypothesized core sleep requirement. The present findings suggest that core sleep might best be considered as the minimum amount of sleep needed by the brain to achieve a state of equilibrium in which alertness and performance are maintained at a stable but lower-than-normal level. In this view, sleep durations that do not satisfy the core sleep requirement would, across days, result in continued degradation of alertness and performance; whereas sleep durations that satisfy the core requirement would produce deficits in alertness and performance relative to baseline, but degradation would not continue across days indefinitely – an asymptotic, stable level of reduced alertness and performance would eventually be achieved; and additional sleep (i.e. incremental increases in the duration of sleep beyond the core requirement) would produce correspondingly higher, and stable, levels of alertness and performance.

Performance stabilization during, and failure to recover following, sleep restriction – evidence for neuromodulatory changes

  1. Top of page
  2. SUMMARY
  3. References

Following chronic, mild to moderate sleep restriction (5 or 7 h TIB), 3 days of recovery sleep (8-h TIB) did not restore performance to baseline levels. These findings suggest that in response to chronic sleep restriction the brain undergoes adaptive changes that serve to sustain a stable (albeit reduced) level of performance; the findings further suggest that these changes persist into the recovery period and prevent rapid return to baseline performance. Such adaptive changes may act as a rate limiter or governor that reduces (caps) the operational capacity of the brain, allowing the brain to operate in the face of a restricted sleep budget. They may serve to prevent injury from occurring in the brain if it continued to perform at full capacity in the face of restricted sleep. Plausible hypothetical mechanisms underlying such an adaptive response to sleep restriction include alterations in gene expression. Adaptive alterations in gene expression could, in turn, cause relatively persistent changes in synaptic transmission (neuromodulation) (Siegelbaum et al. 2000). The possibility that acute TSD and chronic sleep restriction could each produce distinct and different patterns of up and down regulation in gene expression is being investigated in our laboratory.

The nature of sleep dose-dependent performance effects

  1. Top of page
  2. SUMMARY
  3. References

It has been hypothesized (e.g. Lubin 1967; Williams et al. 1959) that all sleep-loss-induced performance deficits are the result of ‘lapses’ in performance – perhaps due to brief episodes of EEG-defined sleep, and that performance between lapses (i.e. during EEG-defined wakefulness) may be unaffected by sleep loss. Although this hypothesis has generally and repeatedly been shown to be incorrect (for review, see Dinges and Kribbs 1991), we tested this hypothesis by analyzing the fastest 10% of responses (as this subset of the data would not be impacted by occasional lapses to the same extent as measures of central tendency). However, as expected, this analysis revealed a pattern of results that was comparable with that produced by analysis of mean speed, suggesting that performance degradation during sleep restriction in the present study cannot be explained solely by an increased incidence of lapses. Had speed decrements been restricted only to a few outliers (lapses), it could be argued that the effects were not neuromodulatory but rather were due to an increased frequency of transitory attention lapses against a background of otherwise normal performance. The finding that the fastest 10% of responses (i.e. responses other than lapses) as well as mean speed was affected suggests a persistent, pervasive change in brain function.

Total sleep deprivation vs. chronic sleep restriction

  1. Top of page
  2. SUMMARY
  3. References

The pattern of results from the present study do not support the general assumption that acute TSD and chronic sleep restriction are physiologically equivalent phenomena, differing only in magnitude and time course. The present findings indicate that acute TSD and chronic sleep restriction cause differential performance effects that are most salient during the recovery period. That is, chronic sleep restriction leads to long-time-constant changes that may have adaptive value, serving to stabilize performance. However, these adaptive changes appear to come at a cost – brain operational capacity is capped in a manner that apparently precludes rapid recovery to baseline levels of alertness and performance when sleep durations are extended to baseline levels. In contrast, rapid recovery to baseline is typical of performance following acute TSD.

Sleep latency as a gauge of performance capacity

  1. Top of page
  2. SUMMARY
  3. References

In studies of sleep deprivation and sleep restriction, sleepiness is generally (albeit often implicitly) assumed to be the intervening variable that mediates sleep loss-induced performance deficits. Measures such as the MSLT and maintenance of wakefulness test (MWT) – both of which measure the tendency to initiate sleep – are considered relatively direct ways to gauge the extent of the brain's physiological need for sleep (Carskadon and Dement 1981). In contrast, psychomotor performance (while obviously impacted by sleep loss) is thought to reflect sleepiness less directly because of the relatively wide array of additional, non-sleep-related factors known to impact performance. Therefore, it is interesting to note that in the present study TIB groups were more reliably distinguished by PVT performance than by sleep latency scores. This may have been due in part to sleep latency floor effects and/or relatively reduced statistical power (because SLTs were administered half as often as PVTs). However, it may also have been due to the fact that sleep latency scores were somewhat restricted, with a mean sleep onset latency of 6.3 min (i.e. near the pathological range) on the baseline day (despite spending 8 h TIB on the previous three nights). This finding is consistent with those of Harrison and Horne (1996) who reported that sleep latencies for a large portion of their normal subjects were within the pathological range – prompting them to hypothesize that individuals vary considerably with respect to ‘sleepability’– the capacity to initiate sleep regardless of sleep need. A crucial implication of these findings is that human performance capacity cannot be reliably inferred from sleep latency data and as it is ultimately human performance capacity (not tendency to initiate sleep) that determines safety and productivity in the operational environment as well as quality of life for sleep disordered patients, greater efforts to define fitness for duty and severity of sleep-related disorders on the basis of measured performance capacity are warranted.

Implications for modeling and performance prediction

  1. Top of page
  2. SUMMARY
  3. References

Extant mathematical models predicting alertness and performance from preceding sleep–wake history typically involve three factors – sleep homeostasis, circadian rhythm, and sleep inertia (Akerstedt and Folkard 1997). These three-factor models are relatively successful in predicting the effects of acute TSD on alertness and performance and the time course of subsequent recovery. They are less successful in predicting the effects of chronic sleep restriction on alertness and performance and the time course of subsequent recovery (Van Dongen and Dinges 2002; Van Dongen et al. 2002). It appears that this failure cannot be remedied by better model parameterization. Rather, this failure likely represents a basic structural inadequacy of three-factor models in predicting the effects of chronic sleep restriction. The present findings suggest that incorporation of a fourth factor would substantially improve prediction – a factor (presumably reflecting neuromodulatory changes) representing slow (days, weeks) adaptation to chronic sleep restriction and slow de-adaptation following restoration of normal sleep amounts.

Is there an ‘optimum’ sleep amount?

  1. Top of page
  2. SUMMARY
  3. References

With respect to the question of what constitutes optimum sleep, our findings that performance tracked sleep in a dose-dependent manner across the range of sleep restriction/augmentation suggests that humans benefit from progressively greater nightly amounts of sleep (although the amount of benefit accrued with progressively greater sleep amounts may be asymptotic; Harrison and Horne 1995). Future work will replicate and extend these and other findings in chronic sleep restriction, describe the time course of performance degradation and recovery in both acute TSD and chronic sleep restriction, and characterize individual differences in tolerance to sleep deprivation and sleep restriction. The present results highlight the need for careful attention to the sleep–wake history of subjects for the weeks (or even months) preceding the actual experimental observations. Further studies of sleep restriction will facilitate the development of mathematical models with greater power and generality for the prediction of alertness and performance based on sleep–wake history.

Acknowledgments

  1. Top of page
  2. SUMMARY
  3. References

This work was supported by the US Army Medical Research and Materiel Command and the US Department of Transportation (Contract/Grant No. DTFH61-94-Y-00090). We thank Ms Deborah Freund and Mr Neill Thomas of the US Department of Transportation for their support of this work and the staff of the General Clinical Research Center, Johns Hopkins University School of Medicine, Bayview Campus, for providing a continuously staffed 24-h per day venue for this study. We thank the two anonymous reviewers for their thoughtful and constructive comments on this paper.

This material has been reviewed by the Walter Reed Army Institute of Research, and there is no objection to its presentation and/or publication. The opinions or assertions contained herein are the private views of the authors and are not to be construed as official or as reflecting the position of the Department of the Army or the Department of Defense. This study was approved by the Walter Reed Army Institute of Research Human Use Committee and the United States Army Medical Research and Materiel Command Human Subjects Review Board of the Army Surgeon General and was performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki.

References

  1. Top of page
  2. SUMMARY
  3. References
  • Achermann, P., Dijk, D. J., Brunner, D. P. and Borbely, A. A. A model of human sleep homeostasis based on EEG slow-wave activity: quantatative comparison of data and simulations. Brain Res. Bull., 1993, 31: 91113.
  • Akerstedt, T. and Folkard, S. Predicting duration of sleep from the three-process model of alertness regulation. Occup. Environ. Med., 1996, 53: 136141.
  • Akerstedt, T. and Folkard, S. Validation of the S and C components of the three-process model of alertness regulation. Sleep, 1995, 18: 16.
  • Akerstedt. T. and Folkard, S. The three-process model of alertness and its extension to performance, sleep latency, and sleep length. Chronobiol. Int., 1997, 14: 115123.
  • Balkin, T., Thorne, D., Sing, H., Thomas, M., Redmond, D., Wesensten, N., Russo, M., Williams, J., Hall, S. and Belenky, G. Effects of Sleep Schedules on Commercial Motor Vehicle Driver Performance. Report MC-00–133, National Technical Information Service, U.S. Department of Transportation, Springfield, VA, 2000.
  • Bliwise, D. L., King, A. C., Harris, R. B. and Haskell, W. L. Prevalence of self-reported poor sleep in a healthy population aged 50–65. Soc. Sci. Med., 1992, 34: 4955.
  • Bonnet, M. H. and Arand, D. L. We are chronically sleep deprived. Sleep, 1995, 18: 908911.
  • Borbely, A. A., Achermann, P., Trachsel, L. and Tobler, I. A two-process model of sleep regulation. Hum. Neurobiol., 1982, 1: 195204.
  • Borbely, A. A. et al. Sleep initiation and initial sleep intensity: interactions of homeostatic and circadian mechanism. J. Biol. Rhythms., 1989, 4: 149160.
  • Carskadon, M. A. and Dement, W. C. Cumulative effects of sleep restriction on daytime sleepiness. Psychophysiology, 1981, 18: 107113.
  • Coren, S. Sleep Thieves. Free Press, New York, 1997.
  • Dinges, D. F. and Kribbs, N. B. Performing while sleepy: effects of experimentally-induced sleepiness. In: T. H.Monk (Eds) Sleep, Sleepiness, and Performance. John Wiley & Sons, Chichester, UK, 1991: 97128.
  • Dinges, D. F., Maislin, G., Kuo, A., Carlin, M. M., Powell, J. W., Van Dongen, H. and Mullington, J. Chronic sleep restriction: neurobehavioral effects of 4, 6 and 8 h TIB. Sleep, 1999, 22 (1 Suppl.): 115.
  • Dinges, D. F., Pack, F., Williams, K., Gillen, K. A., Powell, J. W., Ott, G. E., Aptowica, C. and Pack, A. I. Cumulative sleepiness, mood disturbance, and psychomotor vigilance performance decrements during a week of sleep restricted to 4–5 hours per night. Sleep, 1997, 20: 267277.
  • Dinges, D. F. and Powell, J. W. Microcomputer analyses of performance on a portable, simple, visual RT task during sustained operations. Behav. Res. Meth. Instrum. Comput., 1985, 17: 652655.
  • Drummond, S. P., Brown, C. G., Gillin, J. C., Stricker, J. L., Wong, E. C. and Buxton, R. B. Altered brain response to verbal learning following sleep deprivation. Nature, 2000, 403: 655657.
  • Folkard, S. and Akerstedt, T. A. Three-process model of the regulation of alertness and sleepiness. In: R.Ogilvie and R.Broughton (Eds) Sleep, Arousal and Performance: Problems and Promises. Birkhauser, Boston, 1991: 1126.
  • Friedman, J., Globus, G., Huntley, A., Mullaney, P., Naitoh, P. and Johnson, L. Performance and mood during and after gradual sleep reduction. Psychophysiology, 1977, 14: 245250.
  • Harrison, Y. and Horne, J. A. ‘High sleepability without sleepiness’. The ability to fall asleep rapidly without other signs of sleepiness. Neurophysiol. Clin., 1996, 26: 1520.
  • Harrison, Y. and Horne, J. A. Should we be taking more sleep? Sleep, 1995, 18: 901907.
  • Hoddes, E., Zarcone, V. P., Smythe, H., Phillips, R. and Dement, W. C. Quantification of sleepiness: a new approach. Psychophysiology, 1973, 10: 431436.
  • Horne, J. A. Why We Sleep. Oxford University Press, New York, 1988.
  • Horne, J. A. and Wilkinson, S. Chronic sleep reduction: daytime vigilance performance and EEG measures of sleepiness with particular reference to ‘practice’ effects. Psychophysiology, 1985, 22: 6978.
  • Jewett, M. E., Dijk, D. J., Kronauer, R. E. and Dinges, D. F. Dose–response relationship between sleep duration and human psychomotor vigilance and subjective alertness. Sleep, 1999, 22: 171179.
  • Jewett, M. E. and Kronauer, R. E. Interactive mathematical models of subjective alertness and cognitive throughout in humans. J. Biol. Rhythms, 1999, 14: 588597.
  • Kirk, R. E. Experimental Design: Procedures for the Behavioral Sciences, 3rd edn. Brooks/Cole, Pacific Grove, California, 1995.
  • Kripke, D. F., Garfinkel, L., Wingard, D. L., Klauber, M. R. and Marler, M. R. Mortality associated with sleep duration and insomnia. Arch. Gen. Psychiatry, 2002, 59: 137138.
  • Kuo, A., Carlin, M., Powell, J. and Dinges, D. Chronic restriction of sleep to 4 hours per night for 14 nights changes performance linearly but not subjective sleepiness. Sleep, 1998, 21 (3 Suppl.): 241.
  • Leger, D. The cost of sleep-related accidents: a report for the. Natl. Commission Sleep Disord. Res. Sleep, 1994, 1994, 17: 8493.
  • Lubin, A. Performance under sleep loss and fatigue. In: S. S.Kety, E. V.Evarts and H. L.Williams (Eds). Sleep and Altered States of Consciousness. Williams & Wilkins, Baltimore, 1967: 506513.
  • Mitler, M. M., Carskadon, M. A., Czeisler, C. A., Dement, W. C., Dinges, D. F. and Graeber, R. C. Catastrophes, sleep, and public policy. Consensus Rep. Sleep, 1988, 11: 100109.
  • Mullaney, D. J., Johnson, L. C., Naitoh, P., Friedmann, J. K. and Globus, G. G. Sleep during gradual sleep reduction. Psychophysiology, 1977, 14: 237244.
  • National Sleep Foundation. Sleep in America Poll. http://www.sleepfoundation.org.
  • Palinkas, L. A., Suedfeld P. and Steel G. D. Psychological functioning among members of a small polar expedition. Avia. Space Environ. Med., 1995, 66: 943950.
  • Pilcher, J. J. and Huffcutt, A. I. Effects of sleep deprivation on performance: a meta-analysis. Sleep, 1996, 19: 318326.
  • Rechtschaffen, A. and Kales, A. A Manual of Standardized Techniques and Scoring System for Sleep Stages of Human Subjects. Brain Information Service and Brain Research Institute, Los Angeles, CA, 1968.
  • Siegelbaum, S. A., Schwartz, J. H. and Kandel, E. R. Modulation of synaptic transmission: second messengers. In: E. R.Kandel, J. H.Schwarz and T. M.Jessell (Eds) Principles of Neural Science. McGraw-Hill, New York, NY, 2000: 229252.
  • Spiegel, K., Leproult, R. and Van Cauter, E. Impact of sleep debt on metabolic and endocrine function. Lancet, 1999, 354: 14351439.
  • Van Dongen, H. P. A. and Dinges, D. F. Chronic partial sleep deprivation data point to a novel process regulating waking behavioural alertness. Sleep, 2002, 11 (Suppl. 1): 232.
  • Van Dongen, H. P. A., Shah, A. D., De Brunier, A. B. and Dinges, D. F. Behavioural alertness and the two-process model of sleep regulation during chronic partial sleep deprivation. Sleep, 2002, 11 (Suppl. 1): 233.
  • Webb, W. B. and Agnew, H. W. The effects of chronic limitation of sleep length. Psychophysiology, 1974, 11: 265274.
  • Webb, W. B. and Agnew, H. W. Are we chronically sleep deprived? Bull. Psychonomic Soc., 1975, 6: 4748.
  • Wesensten, N. J., Balkin, T. J. and Belenky, G. Does sleep fragmentation impact recuperation? A review and reanalysis. J. Sleep Res., 1999, 8: 237245.
  • Williams, H. L., Lubin, A. and Goodnow, J. J. Impaired performance with acute sleep loss. Psychol. Monographs, 1959, 73: 126.