Sleep health dimensions and shift work as longitudinal predictors of cognitive performance in the UK Biobank cohort.
Ell J., Schiel JE., Feige B., Riemann D., Nyhuis CC., Fernandez-Mendoza J., Vetter C., Rutter MK., Kyle SD., Spiegelhalder K.
STUDY OBJECTIVES: The long-term effects of sleep health and shift work on cognitive performance are unclear. In addition, research has been limited by small sample sizes and short follow-up periods. We conducted one of the largest examinations of the longitudinal influence of sleep health dimensions and shift work on cognitive performance in people of middle and old age using data from the UK Biobank. The hypothesis was that poor sleep health and shift work would predict lower cognitive performance. METHODS: Self-reported sleep duration, daytime sleepiness, insomnia symptoms, chronotype and shift work status were assessed as predictors at baseline. Cognitive performance was operationalised by a touchscreen test battery at follow-up between 7.4 ± 2.2 and 9.0 ± 0.9 years after baseline assessment, depending on the specific task. Models were performed for each cognitive domain including relevant confounders (e.g., depression). The alpha level was set at p<0.01 for all analyses. RESULTS: The study sample comprised 9,394 participants for the reasoning task, 30,072 for the reaction time task, 30,236 for the visual memory task, 2,019 for the numeric memory task and 9,476 for the prospective memory task. Shift work without night shifts (ß= -2.0x10- 1 ± 6.5x10- 2, p=0.002) and with night shifts (ß= -1.9x10- 1 ± 7.2x10- 2, p=0.010) predicted a significantly reduced performance in the reasoning task. Short sleep duration (ß= -2.4x10- 1 ± 7.9x10- 2, p=0.003) and shift work without night shifts (ß= -3.9x10- 1 ± 1.2x10- 1, p=0.002) predicted a significantly lower performance in the task probing prospective memory. CONCLUSIONS: Our results suggest that, after controlling for confounding variables, shift work and short sleep duration are important predictors for cognitive performance in people of middle and old age. Further work is required to examine causal mechanisms of the observed associations.