Sleep structure from activity data in rotational shift workers

Abstract

Introduction. Sleep physiology is difficult to capture outside the sleep laboratory. Through non-linear conversion of wrist movement to “locomotor inactivity during sleep” (LIDS), real-life actimetry data can be used to expose movement patterns during sleep that directly reflect ultradian sleep structure. In previous work, we have demonstrated sex-, age-, and work-related (e.g., shift work vs. daywork) differences in LIDS, but a detailed analysis in rotational shift workers is missing. Methods. Wrist-actimetry data was collected over 5 weeks in 15 male shift workers (age 33  8 years, range 22 - 47), working a 3 x 8-h shift schedule with morning (M, starting at 6:00), evening (E, 14:00), and night (N, 22:00) shifts. Mixed-effects regression models were used to determine the (interaction) effects of shift type (M, E, N), chronotype (MSFEsc, midsleep timing on work-free days after E shifts), and number of consecutive shifts (2 vs. 4) on LIDS levels and decline rates, adjusting for age, 24-h sleep duration, and time awake. Results. Shift type: Compared with the evening shift, LIDS (= inactivity) levels for sleep before the morning shift were numerically lower (bM = -1.14) and declined more slowly across ultradian cycles (bM*Cycle = 0.96, all n.s.). In contrast, LIDS levels for sleep after the night shift were numerically higher (bN = 0.97) and declined faster (bN*Cycle = -0.13, all n.s.). Shift type*Chronotype: No significant interaction effects with chronotype were seen for morning and evening shifts, but more sleep during the previous 24 hours decreased LIDS levels significantly after evening shifts (b24-SD = -1.66, p = 0.03). For sleep after night shifts, LIDS levels were lower (bMSFEsc = -1.65, n.s.) and declined significantly slower (bMSFEsc*Cycle = 0.70, p = 0.04), the later the individual’s chronotype. Shift type*Chronotype*Shift number: Compared with 4 consecutive night shifts, LIDS levels were higher and declined faster for later chronotypes when working 2 night shifts in a row. Conclusions. LIDS is a promising tool to study sleep physiology in the real world, including operationally complex environments. LIDS reflected shift- and chronotype-specific differences in sleep: lower levels and slower decline rates were observed for later chronotypes on night shifts, indicating less homeostatic sleep pressure compared with earlier chronotypes. Higher LIDS levels and faster decline rates in late chronotypes when working 2 compared with 4 consecutive night shifts were in line with previously published results in the same sample, showing an increased ‘sleep debt’ from preceding morning shifts.