The prevalence of obesity has increased unremittingly in most countries in recent decades (4). Numerous factors have fuelled this trend, but diet timing has garnered little research attention. In this study we show that adults who consume the majority of daily caloric intake closer to mid-sleep time are more likely to be overweight or obese, even after inclusion of daily caloric intake as an adjustment. Preclinical studies have shown obesogenic effects of rest phase feeding in rodents (290, 291), but a clear limitation of these studies is the use of other animals: it would be premature to extrapolate that the same would be seen in humans. A small initial study of 52 adults with intermediate and late chronotypes reported that calorie consumption after 20:00 was correlated with BMI (246), but the first study to measure diet timing relative to circadian phase reported that
dietary phase angle but not clock time of caloric intake independently associates with adiposity in a group of 110 young adults (268). Our study supports the findings of the latter study and builds on it by including a larger group of adults.
In contrast to some prior research (268), we found that calorie consumption closer to sleep onset was associated with longer sleep. The reason for this discrepancy between studies is unclear. We used different statistical methods to the previous work, and another plausible explanation for our divergent results is that we only considered the nocturnal sleep bout, whereas McHill and colleagues also considered naps (268). On one hand, diet-induced thermogenesis might be expected to offset the decline in body temperature that accompanies sleep onset, and pre-sleep caloric intake might thereby disrupt sleep. On the other, sufficient energy availability may be necessary to sustain sleep and prevent premature awakening to acquire food. Both studies only considered caloric intake timing, but the energy content and nutrient profile of the final dietary event are also likely to influence sleep. Experimental manipulation of diet composition and timing relative to sleep onset will rectify contradictory results of these studies.
Digital means of recording diet have produced higher resolution insights into diet timing of late (266-268). Like the two of these studies carried out in the West (266, 268), we report that participants began consuming calories later on the weekends, and participants also consumed breakfast and lunch later. We also plotted macronutrient intakes according to time of day, showing subtle differences between weekdays and weekends. Unlike previous studies, a clear strength of our work is the use of a diet recall tool that has both been validated against biomarkers of dietary intakes and was developed explicitly for use with the population from which participants came.
As later sleep timing has been associated with later meal timing (246), we anticipated that later mid-sleep time would be associated with a later caloric midpoint. We also hypothesised that sleep period would be inversely associated with caloric period. The data supported our contentions. Interestingly, however, we did not find that more variable sleep timing was related to more variable diet timing. Perhaps eating schedules are more dependent on cultural norms than individual variation in sleep timing. Alternatively, our study may have been underpowered to detect subtle associations between sleep timing variability and fluctuating diet timing.
Our study has notable strengths, including use of relatively accurate sleep measurement devices, use of a validated diet recall tool, and a larger participant sample than similar previous research (246, 268). Nevertheless, we acknowledge that our work has limitations. Few nights of sleep were measured for each participant, and no weekend nights were included. We also did not collect information about recent trans-meridian
travel or use of alarms, and participants did not complete sleep logs to verify time in bed. Furthermore, we did not directly measure circadian phase.
More experiments are necessary to clarify the metabolic and behavioural consequences of manipulating diet timing relative to circadian phase, and we can only speculate about the best measures of diet timing at present. It is plausible that foods consumed later in the waking day are less nutritious, and this could contribute to associations between dietary phase angle and body composition. Additional studies will benefit from precise methods of body composition measurement, like dual-energy X-ray absorptiometry. Furthermore, circadian phase is not fixed. If sleep timing and circadian phase shift over the course of the working week, should diet timing shift in lockstep? This is a particularly pertinent question for rotating shift workers and frequent flyers. It has been shown that diet-induced thermogenesis is lower late than early in the biological day (423), but effects of diet timing on key determinants of body composition such as substrate oxidation and skeletal muscle protein synthesis are not well characterised. Exactly how diet influences components of physical activity such as non-exercise activity thermogenesis is also unclear. Furthermore, experiments using polysomnography are required to show how diet composition and timing interact to influence sleep parameters.
Our study adds credence to the notion that when we eat may be a critical determinant of our metabolic health. Should other studies continue to support this idea, it would make sense to include advice on diet timing in dietary guidelines. Perhaps diet timing is another piece in the complex puzzle we face in preventing and reversing obesity.
Chapter 4: Sleep duration, nutrient intakes, and metabolic health in UK adults: