The impact of sleep timing and bright light exposure on attentional impairment during night work

Abstract

The prevalence of hazardous incidents induced by attentional impairment during night work and ensuing commute times is attributable to circadian misalignment and increased sleep pressure. In a 10-day shift work simulation protocol (4 day shifts and 3 night shifts), the efficacies of 2 countermeasures against nighttime (2300 to 0700 h) attentional impairment were compared: (1) Morning Sleep (0800 to 1600 h; n = 18) in conjunction with a phase-delaying light exposure (2300 to 0300 h), and (2) Evening Sleep (1400 to 2200 h; n = 17) in conjunction with a phase-advancing light exposure (0300 to 0700 h). Analysis of the dim light salivary melatonin onset indicated a modest but significant circadian realignment in both sleep groups (evening sleep: 2.27 +/- 0.6 h phase advance, p < 0.01; morning sleep: 4.98 +/- 0.43 h phase delay, p < 0.01). Daytime sleep efficiency and total sleep time did not differ between them or from their respective baseline sleep (2200 to 0600 h; p > 0.05). However, on the final night shift, the evening sleep subjects had 37% fewer episodes of attentional impairment (long response times: 22 +/- 4 vs. 35 +/- 4; p = 0.02) and quicker responses (p < 0.01) on the Psychomotor Vigilance Task than their morning sleep counterparts. Their response speed recovered to near daytime levels (p = 0.47), whereas those of the morning sleep subjects continued to be slower than their daytime levels (p = 0.008). It is concluded that partial circadian realignment to night work in combination with reduced homeostatic pressure contributed to the greater efficacy of a schedule of Evening Sleep with a phase-advancing light exposure as a countermeasure against attentional impairment, over a schedule of Morning Sleep with a phase-delaying light exposure. These results have important implications for managing patients with shift work disorder.

Figure 1. Double Raster Plots of the Shift Work Protocol

The x-axis represents clock time and the y-axis represents the day of the study. Hashed bars represent work shifts (day shift: 0700 – 1500; night shifts: 2300 – 0700). The open bars on the night shifts represent light exposure (Evening Sleep: 2300 – 0300; Morning Sleep: 0300 – 0700). The black bars represent the sleep episodes (Baseline Sleep: 2200-0600; Morning Sleep: 0800 –1600; Evening Sleep: 1400 – 2200). Time spent outside the laboratory was from 1530 – 2100 on day shifts and on night shifts it was from 0730-1300 for the Evening Sleep group and from 1630 – 2200 for the Morning Sleep group. The gray bars represent the constant posture (CP) episode and constant routine (CR).

Figure 2. DLSMO_25% in the Evening and Morning Sleep groups

These data represent the individual DLSMO_25% (filled circles), and average (mean ± SEM) DLSMO_25% (open circles) and average CBT_min (inverted triangles) for the four sleep × light intensity groups. The x-axis represents clock time and the y-axis represents the endogenous circadian phase assessments. The horizontal hashed bars represent the night shifts and the open bars the timing of the Light Exposure respectively. In the Evening Sleep group, 7 out of 8 subjects in the 600 lux condition and all subjects in the 2500 lux (n = 8) condition exhibited a phase advance in DLSMO_25%. One subject (gray circle, top left panel) in the Evening Sleep, 600 lux condition exhibited 4:18 h phase delay and another a 12 minute phase delay. In the Morning Sleep group, all subjects in the 600 lux (n = 9) and 2500 lux (n = 8) conditions exhibited a phase delay. CBT_min in the Evening Sleep subjects (600 lux: 1:13 ± 1:12 h; 2500 lux: 23:46 ± 1:02 h) was phase advanced relative to the Morning Sleep subjects (600 lux: 8:43 ± 0:59 h; 2500 lux: 8:24 ± 0:59 h).

Figure 3. PVT Cumulative Response Time Distribution

The data represent the average RT percentiles (Mean ± SEM; symbols) and the fitted CDF (4-parameter Weibull) for the four work shifts. The data for each sleep group (Morning Sleep: n = 17; Evening Sleep: n = 16) is averaged over the light exposure conditions. The PVT was administered every two hours starting at 0830 on the day shifts and at 0030 on the night shifts. The bottom x-axis represents Log transformed RT and the top x-axis the equivalent RT in milliseconds. The y-axis represents percentile points. A rightward shift along the x-axis corresponds to a lengthening of RT and a stretching of the tail of the distribution corresponds to an increase in long RT. The ‘∼’ represents a comparison with the day shift RT curve (F-test) at p = 0.05 level.

Figure 4. Episodes of Attentional Impairment and Subjective Alertness

The top panel shows the number of long RTs (Mean ± SEM) per work shift for the two sleep groups (Morning Sleep: n = 17; Evening Sleep: n = 16) averaged over the light exposure conditions. The x-axis represents work shift and the y-axis represents the number of RT outliers. Increasing values reflect increasing attentional impairment. The bottom panel shows the KSS scores (Mean ± SEM) per work shift for the two sleep groups (Morning Sleep: n = 17; Evening Sleep: n = 16) averaged over the light exposure conditions. The x-axis represents work shift and the y-axis represents the KSS score. Increasing values reflect deteriorating alertness. The KSS was administered every two hours starting at 0830 on the day shifts and at 0030 on the night shifts. The symbols ‘*’ and ‘∼’ represent a comparison to baseline sleep at the p < 0.01 and p < 0.05 level respectively (paired t-test).

Figure 5. TST and REM Sleep on Day Sleep 3

Distribution of TST (left panel) and REM sleep (right panel) across thirds of Day Sleep 3. Morning Sleep group is indicated by black bars (n = 14) and Evening Sleep group by gray bars (n = 16). Data represent the means + SEM.