The impact of experimentally shortened sleep on timing of eating occasions in adolescents: A brief report

Summary

Short sleep increases the risk for obesity in adolescents. One potential mechanism relates to when eating occurs in the day. This study investigated the impact of shortened sleep on eating occasion timing in adolescents. Ninety-three healthy 14- to 17-year-olds (62% female) completed a within-subject experimental sleep manipulation, engaging in 5-night spans of Short Sleep (6.5-hr sleep opportunity) or Healthy Sleep (9.5-hr sleep opportunity), with order randomized. During each condition, adolescents completed three 24-hr diet recall interviews. Repeated-measure t-tests assessed the sleep manipulation effect on each adolescent’s number of meals, first and last eating occasion (relative to the clock and time since sleep onset/offset), feeding window (timespan from first to last eating), and the midpoint of feeding. The timing of the first eating occasion was similar across conditions, relative to the clock (Short = 08:51, Healthy = 08:52) and to time since waking (Short = 2.0 hr, Healthy = 2.2 hr). The timing of the last eating occasion was later relative to the clock (Short = 20:34, Healthy = 19:39; p < 0.001), resulting in a longer feeding window (Short = 11.7 hr, Healthy = 10.8 hr, p < 0.001) and a later midpoint in the feeding window (Short = 14:41, Healthy = 14:18, p = 0.002). The gap between last eating occasion and sleep onset was larger in Short (4.2 hr) than Healthy Sleep (2.9 hr; p < 0.001), though the last eating occasion was much earlier than when they fell asleep during either condition. Shortened sleep resulted in adolescents eating later and lengthening the daily feeding window. These findings may help explain the link between shortened sleep and increased obesity risk in adolescents.

1 |. INTRODUCTION

Most adolescents regularly obtain inadequate sleep, a phenomenon deemed a public health crisis (Seton & Fitzgerald, 2021). Poor sleep has been linked with increased risk for several physical and mental health disorders (Illingworth, 2020), including increased obesity risk in adolescents (Duraccio et al., 2019). The mechanisms underlying the relationship between poor sleep and increased obesity risk remain unclear, but there is evidence that sleep restriction worsens dietary quality (Duraccio et al., 2019), particularly in the evening (Duraccio et al., 2022). In youth, eating later in the day and across a longer span of time during the day is associated with increased risk for obesity (Eng et al., 2009), and adults undergoing experimental sleep restriction have been shown to shift their meal timing to later hours (Spaeth et al., 2013). The effects of sleep restriction on the timing of eating occasions in adolescents remains unknown but may be an important factor, given that adolescent obesity rates and sleep deprivation are on the rise (Hales et al., 2017; Wheaton et al., 2018).

To fill this gap, we ran secondary analyses on a previously described adolescent sleep manipulation study to generate a novel comparison of the timing of eating occasions (e.g. time of first eating occasion, time of last eating occasion, feeding window) across 5 nights of Short Sleep (6.5-hr sleep opportunity) versus 5 nights of Healthy Sleep (9.5-hr sleep opportunity). We hypothesized that, while in Short Sleep, adolescents would consume more meals and have their final eating occasion later in the day, compared with Healthy Sleep. Additionally, we hypothesized that adolescents in Short Sleep would consume foods across a longer window and have their midpoint of feeding later during the day, compared with Healthy Sleep. Finally, we explored the potential moderating impact of several between-subject variables (demographics, body mass index [BMI] z-score, and midpoint of sleep immediately prior to randomization) on sleep condition effects.

2 |. METHODS

2.1 |. Participants

Healthy adolescents aged 14–17 years were invited to participate in this study; further details regarding recruitment and eligibility are detailed in Appendix S1 and in Duraccio et al. (2022), which focused on overall dietary amount and quality, not timing.

2.2 |. Protocol

The Institutional Review Board at Cincinnati Children’s Hospital Medical Center approved and oversaw all study procedures. As detailed in Duraccio et al. (2022), all adolescents engaged in a 3-week repeated-measures sleep experimental protocol, with order of the experimental conditions randomized. Across the entire 3-week protocol, participants selected and maintained a wake-time that would allow them to attend an 08:00 hours meeting. During the first 5 nights of participation, sleep was stabilized (adolescents selected a bedtime to allow for 8 hr of sleep) and adherence to this assigned wake-time was determined. During weekends following this sleep stabilization period and between the experimental conditions, participants were allowed to self-select their bedtimes. Following the stabilization week, adolescents were randomized to either Short Sleep (with bedtime shifting later to allow for a 6.5-hr sleep opportunity) or Healthy Sleep (with a bedtime shifting earlier to allow for a 9.5-hr sleep opportunity). This experimental approach was selected to maximize the generalizability of our findings to the school year and based on pragmatic considerations (altering bedtime is the only practical option for most adolescents to change sleep duration during school nights). Following completion of the first experimental sleep condition, participants crossed over to the other condition after a 2-night washout. Participants completed three randomly assigned dietary assessments during each sleep condition.

2.3 |. Measures

2.3.1 |. Participant demographics

Participants or their parents reported family income, participant age, sex and race (using NIH categories).

2.3.2 |. Body mass index

Height and weight were measured in triplicate by trained research assistants on a calibrated hospital scale. Data were transformed into age- and sex-normed BMI z-scores using US Centers for Disease Control and Prevention norms.

2.3.3 |. Sleep adherence

Sleep adherence was measured using wrist-worn actigraphs and cross-checked for artefacts by reviewing actigraphy data with participants relative to nightly sleep diaries (Duraccio et al., 2022). Actigraphy was scored using the Sadah algorithm (Sadeh et al., 1994), with time in bed being used to determine adherence to the experimental protocol.

2.3.4 |. Dietary outcomes

Across each 5-day experimental condition, participants were called via telephone on three randomized days and asked to complete a 24-hr dietary recall interview by a member of the institution’s bionutrition research core who was blinded to randomization. Participants were asked to recount all foods and beverages consumed on the previous day using a multiple-pass approach, which records content, amount and timing of each eating occasion. For the present analyses, we focused on seven key outcomes: total number of meals, timing of first eating occasion, time between waking and first eating occasion, timing of last eating occasion, time between last eating occasion and sleep onset, feeding window (time from first to last eating occasion), and the midpoint of feeding window.

2.4 |. Data analytic plan

To determine the impact of experimental sleep condition (Short versus Healthy Sleep) on each of the six timing outcomes, we conducted a series of repeated-measure t-tests using SPSS (Version 28). We then explored the moderating impact of sex, age, BMI z-score, income, race and midpoint of baseline sleep on sleep condition differences in dietary outcomes using a series of repeated-measures generalized linear models. Significance thresholds for hypothesis-driven and exploratory analyses were p = 0.05 and 0.01, respectively. Means ± standard deviations are reported.

3 |. RESULTS

A total of 93 adolescents completed all study procedures and were adherent to the sleep experimental protocol (see Duraccio et al., 2022 and Appendix S1 for information regarding those who were excluded, non-adherent or dropped from the study). Demographics of the final sample are detailed in Table 1 of Appendix S1; differences in sleep outcomes across experimental conditions are summarized in Table 1.

TABLE 1.

Differences in timing of meals and sleep parameters across short and healthy sleep in adolescents

Outcome	Short sleep M (SD)	Healthy sleep M (SD)	p	Cohen’s d
Number of meals consumed	4.13 (1.03)	4.07 (1.17)	0.440	−0.07
First meal timing, relative to clock	08:51 (1:31)	08:52 (1:44)	0.911	0.01
First meal timing, relative to sleep offset (hr)	1.95 (1.60)	2.15 (2.15)	0.355	0.10
Last meal timing, relative to clock	20:34 (1:34)	19:39 (1:18)	< 0.001	−0.49
Last meal timing, relative to sleep offset (hr)	4.15 (1.65)	2.88 (1.43)	< 0.001	−0.63
Range of dietary consumption (hr)	11.72 (2.28)	10.77 (2.12)	< 0.001	−0.37
Midpoint of dietary consumption, relative to clock	14:41 (1:08)	14:18 (1:05)	0.002	−0.30
Sleep onset (time)	00:44 (0:31)	22:30 (0:43)	< 0.001	3.57
Sleep offset (time)	07:01 (0:21)	06:58 (0:27)	0.188	0.12
Sleep period (hr)	6.27 (0.52)	8.47 (0.64)	< 0.001	0.73

As detailed in Table 1 and illustrated in Figure 1, the average time of the first daily eating occasion was similar across Short Sleep and Healthy Sleep relative to the clock (08:51 ± 1:31, 08:52 ± 1:45, respectively; p = 0.911) and to hours since wake onset (1.95 ± 1.60; 2.15 ± 2.15, respectively; p = 0.355). In contrast, adolescents in the Short Sleep condition ate nearly ~1 hr later into the evening (20:34 ± 1:34) compared with when they were in the Healthy Sleep condition (19:39 ± 1:18, p < 0.001), though both were well before sleep onset in either condition. As a result, adolescents had a longer feeding window (Short = 11.72 ± 2.28 hr; Healthy = 10.77 ± 2.12 hr; p < 0.001) and a later midpoint of feeding (Short = 14:41 ± 1:08; Healthy =14:18 ± 1:05; p = 0.002) while in the Short Sleep condition than when in the Healthy Sleep condition. Finally, adolescents in Short Sleep had their final eating occasion 1 hr farther away from their sleep-onset time than when in the Healthy Sleep (4.15 ± 1:65 hr, 2.88 ± 1.43 hr, respectively; p < 0.001). There were no significant differences in the number of meals consumed across the day (p = 0.440). None of the exploratory moderators was statistically significant (p >0.075).

FIGURE 1.

Overview of daily meal timing across experimental sleep conditions, relative to sleep patterns observed. With dark grey indicating patterns observed in short sleep and light grey indicating patterns observed in healthy sleep, we observe that adolescents engaging in short sleep are eating later into the evening and for a longer feeding window compared with when engaging in healthy sleep. However, the averaged last meal during short sleep is ~2 hr before the adolescent would have initiated sleep in the healthy sleep condition, indicating that these patterns observed are more than a function of having increased opportunity to eat in the late evening.

4 |. DISCUSSION

The primary aim of this study was to explore how meal timing differed when adolescents obtained experimentally induced Short Sleep experimentally induced Short Sleep compared with Healthy Sleep. Adolescents in Short Sleep had their final eating occasion later in the day, though farther from sleep onset, compared with those in Healthy Sleep. Additionally, we found that adolescents consumed food across a longer daily window and had a later midpoint of consumption during the day during Short Sleep, compared with Healthy Sleep. There were no differences in the timing of the first eating occasion or in number of meals across the experimental conditions.

Staying up later provided more opportunities to eat into the late evening, but that does not seem to explain current findings. The number of meals did not differ across conditions and, although delayed, the final eating bout during the Short Sleep condition was still well before sleep-onset time during the Healthy Sleep condition, both on average (Figure 1) and in > 94% of individual participants. When staying up late, adolescents did not simply eat at times that they otherwise would have been asleep. Instead, they shifted the timing of their meals later, extended their feeding window, and delayed the midpoint of their daily feeding within normal waking hours. The reasons for this shift are not clear. Prior findings suggest increased intake of high-sugar foods later during sleep restriction (Duraccio et al., 2022), so extending eating later into the evening may reflect either an attempt to maintain energy or decreased ability to inhibit in the presence of appealing high-sugar foods (Duraccio et al., 2019).

Regardless of aetiology, the shift towards later meals and a wider and later feeding window could increase the risk for adverse cardiometabolic outcomes. Adults who consume meals in the late evening or night have slowed digestion, leading to increased fat storage and body mass (Garaulet & Gómez-Abellán, 2014) and risk for cardiometabolic disorders (Dashti et al., 2021). Furthermore, longer feeding windows in adults are associated with increased insulin resistance, greater BMI, higher blood pressure, and higher levels of cholesterol, triglycerides and fasting plasma glucose (Patikorn et al., 2021). Though less examined in adolescents, there is some evidence that late evening eating in youth results in increased obesity risk (Eng et al., 2009). This, paired with the tendency for adolescents to skip breakfast (Ardeshirlarijani et al., 2019; also seen in our data) may increase obesity risk. Furthermore, meal consumption that is even modestly misaligned with circadian sleep phase can be problematic to overall health (Baron et al., 2017); while circadian sleep phases were not directly measured in this study, altering bedtime may have resulted in a shift in sleep phase that disrupted the synchrony in circadian rhythms of sleep and meal timing. Future research on circadian misalignment and meal timing is warranted.

Several limitations should be noted. First, we examined the changes in average eating patterns across three dietary recalls and across conditions; it is possible that intra-individual variability impacts the role of sleep duration on timing of eating occasions. Second, our sleep manipulation was relatively brief; future research should extend this experimental protocol over a longer period to determine whether the effects are enhanced over time. Third, our sleep restriction protocol also impacted timing of sleep, so we are unable to determine if the effects are due to sleep duration or due to altering bedtime; future research should examine the specific role of sleep timing on timing of eating occasions. Finally, although this is one of the largest experimental sleep manipulation studies of adolescents published to date, statistical power and the generalizability of our findings are limited by the sample size and a high proportion of our sample being middle- to upper-socioeconomic class; future research should include more youth with socioeconomic disadvantages.

Supplementary Material

SUPPORTING INFORMATION

Additional supporting information can be found online in the Supporting Information section at the end of this article.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.