Original Article | Published:

Interventions and public health nutrition

The effects of partial sleep deprivation on energy balance: a systematic review and meta-analysis

European Journal of Clinical Nutrition volume 71, pages 614624 (2017) | Download Citation

Abstract

Background/Objectives:

It is unknown whether short sleep duration causatively contributes to weight gain. Studies investigating effects of partial sleep deprivation (PSD) on energy balance components report conflicting findings. Our objective was to conduct a systematic review and meta-analysis of human intervention studies assessing the effects of PSD on energy intake (EI) and energy expenditure (EE).

Subjects/Methods:

EMBASE, Medline, Cochrane CENTRAL, Web of Science and Scopus were searched. Differences in EI and total EE following PSD compared with a control condition were generated using the inverse variance method with random-effects models. Secondary outcomes included macronutrient distribution and resting metabolic rate. Heterogeneity was quantified with the I2-statistic.

Results:

Seventeen studies (n=496) were eligible for inclusion in the systematic review, and 11 studies (n=172) provided sufficient data to be included in meta-analyses. EI was significantly increased by 385 kcal (95% confidence interval: 252, 517; P<0.00001) following PSD compared with the control condition. We found no significant change in total EE or resting metabolic rate as a result of PSD. The observed increase in EI was accompanied by significantly higher fat and lower protein intakes, but no effect on carbohydrate intake.

Conclusions:

The pooled effects of the studies with extractable data indicated that PSD resulted in increased EI with no effect on EE, leading to a net positive energy balance, which in the long term may contribute to weight gain.

Introduction

The current National Sleep Foundation guidelines recommend 7–9 h of sleep per night for adults aged 18–64 years, and state <6 h may compromise health and well-being.1 Some evidence suggests that the average human sleep duration has faced rapid declines over the past century;2 however, other data report no significant changes.3, 4 Social demands of the busy modern lifestyle, including excessive screen time and shift work, are hypothesised to contribute to causing the possible reductions in sleep duration as well as potential circadian misalignment.5, 6, 7 The definition of short sleep duration differs widely across studies conducted in adults, and varies from 4 h8 to <8 h.9 Despite this lack of standardisation, most observational data suggests that it is strongly associated with weight gain and non-communicable diseases,10, 11, 12, 13, 14 yet there remain to be conflicts in the evidence.15, 16

Experimental evidence has examined causality of the possible relationship between short sleep duration and risk of weight gain. Intervention studies examining the effects of partial sleep deprivation (PSD), defined as restricted but not complete elimination of sleep, compared with habitual sleep on energy balance are thus far equivocal. It remains unknown whether PSD is causal in weight gain, and if this arises from an excess energy intake (EI), reduced total energy expenditure (EE) or a combination of the two factors. A recent systematic review of randomised controlled trials (RCTs) suggested that restricted sleep may increase dietary intake; however, the extent of this increase was not quantified.17 Understanding the mechanisms associating weight gain and PSD has the potential to provide a novel target for intervention in weight management. To the best of our knowledge, the literature investigating effects of PSD on EI and EE have not previously been systematically evaluated and quantitatively analysed by meta-analysis. Therefore, we aimed to systematically review and meta-analyse the effects of PSD on components of the energy balance equation including EI and total EE, as well as macronutrient distribution and resting metabolic rate (RMR) in comparison with habitual sleep in healthy adults.

Materials and methods

We conducted a comprehensive search on five databases, Medline, EMBASE, Cochrane CENTRAL, Scopus and Web of Science, searching all years of record until 17 November 2014 with no language restrictions. We also searched reference lists of relevant literature, and for unpublished studies at www.clinicaltrials.gov. Search terms were determined in collaboration with all the authors through exploration of key words in the literature. The search terms included ‘Energy Balance’, or ‘Energy Intake’, or ‘Energy Expenditure’, or ‘Resting Metabolic Rate’ and ‘Sleep Deprivation’ and ‘Sleep Curtailment’. For the complete search strategy, please see the Supplementary Material online. A study protocol was prepared in line with Cochrane18 and PRISMA19 guidelines and registered with International Prospective Register of Systematic Reviews (PROSPERO) (Registration Number: CRD42014014978).

Study inclusion and exclusion criteria

We included studies assessing healthy male and/or female human participants with no diagnosed chronic diseases or sleep conditions of all body mass index (BMI) ranges, aged 18 years. Laboratory-based or free-living interventional studies were included if they assessed the effect of PSD, where sleep duration was shortened compared with habitual sleep, but not eliminated completely, on energy balance. The minimum and maximum number of hours of sleep for the PSD condition was not prespecified, as the literature from observational studies provides no consensus on the definition of short sleep duration. Acute and chronic studies of PSD were considered. Eligible studies were required to include at least one night of normal or habitual sleep opportunity allowing longer duration of sleep than the PSD condition as control or baseline. The primary outcomes of this review were 24-h total EI, ad libitum EI and 24-h total EE. Secondary outcomes included macronutrient (protein, carbohydrate and fat) intakes and RMR.

We excluded total sleep deprivation interventions, as effects of complete elimination of sleep were likely to be exaggerated and unrepresentative of shortened sleep duration. We also excluded fragmented sleep studies as the effects pertain to disrupted sleep quality, and not duration. Studies being conducted alongside any other intervention such as weight loss diets/exercise regimens were excluded as the effects may interfere with the outcomes of interest.

Study designs

Randomised and non-randomised interventional studies of parallel, crossover or ‘pre–post’ study design were considered eligible. ‘Pre–post’ studies assessed changes from baseline, and for these studies we treated the baseline condition as control. We included studies reported in full-text journal articles as well as conference abstracts. Observational studies, reviews, meta-analyses and editorials were excluded.

Data extraction and quality assessment

Two reviewers (HA and either JD, GP or SH) independently screened titles and abstracts of the identified studies for formal review and analysis. The primary author (HA) retrieved full texts for those identified studies and extracted the data. Clarification from an independent reviewer (JD, GP or SH) was sought when required. The data extraction form was designed based on guidelines from Cochrane,18 and included author’s name and year, country, research setting (laboratory or free-living), study design, number of participants, sex, age, BMI, length of treatment, number of hours of sleep in the PSD condition, number of hours of sleep in the control/baseline condition and methods of assessment of sleep, EI and EE. Data from multiple reports of the same study, including journal articles and abstracts, were collated under a single study identifier, and referenced under the main full-text article.

Means and s.d. or s.e.m. for daily EI (kcal or kJ), EE (kcal or kJ) and/or macronutrients for the PSD and control conditions were to be reported for inclusion in the meta-analysis. Where s.e.m. was reported, s.d. was calculated for inclusion in the meta-analysis. Macronutrients were to be reported as a percentage of total energy consumed (%E) to allow for the comparison of macronutrient distribution. When means and s.d. or s.e.m. were not reported (kcal or kJ) in tables or text, as well as when macronutrients were reported in grams, and not %E for macronutrients, attempts were made to contact the corresponding authors for clarification on full-text articles by email, twice in 1 month. If the data were unavailable when authors were contacted, but was presented in a figure, we extracted the data using a visual screen ruler. The primary reviewer (HA) assessed the quality of included studies by using the Cochrane Risk of Bias assessment tool.20

Statistical analysis

Statistical analyses were conducted in Review Manager 5.2 (Copenhagen, Denmark) software. We calculated the raw mean difference, and used the inverse variance method to weight the studies, and pooled data using the random-effects model to quantify difference in means. We tested for between-study heterogeneity using the Q-statistic, and quantified its extent using I2-statistic.21 A sensitivity analysis was performed to separate effects of study design (RCT, non-randomised and pre–post) as prespecified. Publication bias was assessed subjectively by visual inspection of Begg’s funnel plot for skewness.22 A P-value <0.05 was considered statistically significant.

Results

Search results

The details of the literature search are presented in Figure 1. We identified 5843 publications for screening after duplicates were removed. Of these, 86 were retrieved in full text for further in-depth evaluation, from which a total of 28 publications reporting on 16 studies including 496 subjects were identified as suitable for inclusion. Multiple publications reporting on a single study were identified and combined under the main journal article’s first author’s name and year of publication, as outlined in Table 1. Twelve of the publications were full-text articles reporting on 11 studies.23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34 Seven of these studies24, 25, 26, 28, 32, 33, 34 were associated with 11 conference abstracts.35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45 A further five studies were reported in five conference abstracts, with no full-text articles identified,46, 47, 48, 49, 50 which were thus treated as stand-alone abstracts.

Figure 1
Figure 1

Flow chart of literature searching, screening and selection of articles for inclusion in this systematic review on the effect of PSD on energy balance. 1One study (Spaeth et al34) entered twice as means were presented separately for males and females. 2Data extracted from figures using visual screen ruler for two studies (Markwald et al28, Spaeth et al34).

Table 1: Characteristics of included studies assessing the effect of PSD on energy balance

Characteristics of included studies

An overview of the characteristics of studies included in this article is presented in Table 2.

Table 2: Results of studies assessing the effect of PSD on energy balance

Design

Ten randomised crossover studies,24, 25, 27, 29, 32, 46, 47, 48, 49, 50 two non-randomised crossover studies,28, 30 two randomised parallel studies26, 33 and two pre–post design studies23, 34 were included in the present review. The length of the wash-out period in the 10 randomised controlled crossover studies ranged from 5 days51 to 3 months.29 Most studies were conducted in the laboratory setting,24, 26, 27, 28, 29, 32, 33, 34, 46, 49 two were a combination of laboratory setting and free-living,25, 30 and only two were entirely free-living.23, 48 The interventions were generally of a short duration, with <1 week spent in the PSD and control conditions in 14 studies,23, 24, 25, 27, 28, 30, 32, 33, 34, 46, 47, 48, 49, 50 and the longest intervention being 2 weeks per condition.29 Three studies were acute, assessing each condition for only one night.25, 48, 50 The most severe PSD studies restricted sleep duration to 3 h 30 min or < 4 hs per night.47, 50 In the least restrictive interventions, PSD sleep was restricted to 5 h 30 min,29 or to two-thirds of participants’ habitual sleep.26 The method of sleep restriction varied such that participants had different sleep midpoints between studies. Five studies kept the wake point constant and delayed bedtime,26, 27, 30, 33, 34 and five studies shifted wake time and sleep time closer together, evading shifts in sleep midpoints.24, 25, 28, 29, 32 However, six studies did not report bed times and wake times.23, 46, 47, 48, 49, 50 The prescribed time in bed for the control condition varied across studies from 7 h47 to 10 or 12 h.34

Participant characteristics

The 16 studies included 222 males and 244 females, with the sex of the remaining 30 unreported.49, 50 Participants’ age ranged from 18 to 50 years. Of the studies that reported participants’ BMI, four restricted the sample to participants within the normal/healthy BMI range,25, 26, 28, 30 seven included participants within the overweight range23, 24, 27, 29, 32, 33, 34and five studies included obese participants.23, 27, 33, 46, 48 Of the studies that included female participants, only two accounted for the menstrual phase.29, 32

Effect of PSD on EI

Qualitative analysis

Fourteen studies including 480 participants assessed the effect of PSD on EI, of which 12 studies (n=454) reported on mean 24-h EI based on 1 day,24, 25, 30, 48, 49, 50 2–5 days23, 26, 28, 33, 34 or 14 days29 in the PSD and control conditions (As shown in Table 2). The reported dietary assessment methods included dietary records,23 ad libitum food access from inside and outside the laboratory26 or self-selected items from outside the laboratory.52 Four studies used a combination of scheduled meals and ad libitum food access from within the testing facility.28, 29, 33, 34 Four studies included buffet meals for breakfast,30, 50 lunch,25 dinner49 or snack access,30, 49 used in combination with scheduled meals,25, 30 dietary records25 or 24-h recall.50 Two of the aforementioned studies reported using heterogeneous buffet settings including 2025 or >8 items;30 however, two other studies did not state whether the buffet was homogeneous or heterogeneous.49, 50 Two studies reported ad libitum EI from a single buffet breakfast46 or lunch,27 but did not assess total 24-h EI. Of the 12 studies reporting on total 24-h mean EI, seven reported a significantly higher EI in the PSD condition in comparison with the control condition.23, 24, 25, 28, 33, 34, 49 The remaining four studies found no significant difference in EI between the PSD and control conditions,26, 29, 48, 50 of which one was the longest study.29 The two studies reporting EI from a single buffet meal also found no significant difference in EI.27, 46 None of the studies found a reduction in EI in response to PSD compared with the control condition.

Quantitative analysis

Sufficient data were only available to assess quantitatively total 24-h EI. Ten of the 12 studies assessing total 24-h EI were able to be included in the meta-analysis.23, 24, 25, 26, 28, 29, 30, 33, 34, 49 Two were not included as numbers were not reported in the conference abstract.48, 50 The studies included in the meta-analysis comprise of all seven studies reporting an increase in total 24-h EI, and three of the five studies with no effect. Two studies’ data were extracted from figures,28, 34 of which one study34 presented male and female data separately, and was thus entered twice (Spaeth 2014 (m), Spaeth 2014 (f)) into the meta-analysis to account for this. Overall, the quantitative analysis included n=185 and n=161 subjects in the PSD and control conditions, respectively. The pooled mean increase in total 24-h EI in the PSD condition compared with control was 385 kcal (95% confidence interval (CI): 252, 517; P<0.00001), with low heterogeneity (Q=6.28, d.f.=10, P=0.79, I2=0%) (Figure 2). A sensitivity analysis showed that a total 24-h EI was significantly increased as a result of PSD in both the randomised24, 25, 26, 28, 29, 33, 49 (364 kcal (95% CI: 191, 538; P<0.0001)) and non-randomised studies23, 30, 34 (411 kcal (95% CI: 165, 657; P=0.001)) with low heterogeneity (PSD: Q=2.16, d.f.=6, P=0.90, I2=0%; Control: Q=4.00, d.f.=3, P=0.26, I2=25%).

Figure 2
Figure 2

Forest plot of the difference in 24-h total EI changes after PSD in control and intervention groups.

Effect of PSD on EE

Qualitative analysis

Six studies including 87 participants reported on the effect of PSD on total 24-h EE, measured by whole-room indirect calorimetry,28, 32, 47 heart rate monitor23 or doubly labelled water.24, 29 The two studies assessing EE by whole-room indirect calorimetry showed a significant increase (~5%) in EE in response to PSD.28, 32 The four remaining studies found no significant difference in EE as a result of PSD.23, 24, 29, 47

Quantitative analysis

Sufficient data for inclusion in the meta-analysis were available for five of the six studies investigating the effect of PSD on total 24-h EE.23, 24, 28, 29, 32 One study was not included as the numbers were not reported in the conference abstract.47 One study’s data were extracted from a figure.28 The meta-analysis included n=73 PSD and n=73 control participants, as shown in Figure 3. The studies included the two studies observing an increase in total 24-h EE as a result of PSD, and three of the four reporting no effect. There was no significant change in 24-h EE as a result of PSD (88 kcal; 95% CI: −21, 198; P=0.11), with low heterogeneity (Q=1.08, d.f.=4, P=0.90, I2=0%). Exclusion of the single non-randomised study23 did not significantly affect the results (81 kcal; 95% CI: −30, 193; P=0.15).

Figure 3
Figure 3

Forest plot of the difference in 24-h total EE changes after PSD in control and intervention groups.

Effect of PSD on macronutrient intake

Qualitative analysis

Of the 14 studies assessing EI, 12 studies reported on changes in macronutrient intake from mean 24-h EI23, 24, 25, 28, 29, 30, 33, 34, 49, 50 or a single buffet meal.27, 46 The PSD condition was reported to significantly increase fat intake in four studies,24, 25, 30, 34 which one attributed to saturated fat in particular,24 and to reduce protein intake in two studies27, 34 compared with control. Carbohydrate intake was significantly higher in the PSD condition compared with control in two studies.28, 49 However, no significant changes in macronutrient intakes were found in six studies.23, 27, 29, 33, 46, 50 None of the studies reported decreased carbohydrate, decreased fat or increased protein intakes in the PSD condition compared with control.

Quantitative analysis

Of the 10 studies reporting on %E for macronutrient intakes, seven studies provided sufficient data for inclusion in the meta-analyses,23, 24, 25, 29, 30, 33, 34 including n=151 and n=126 subjects in the PSD and control conditions, respectively. Three studies were not included as the numbers were not reported in the conference abstract,49, 50 or data were reported in grams and were unavailable when authors were contacted.28 Fat intake was 1.6%E (95% CI: 0.3, 2.9; P=0.02) higher in the PSD condition compared with control with low heterogeneity (Q=3.45, d.f.=6, P=0.75, I2=0%) (Figure 4a). However, significance was lost when analysed by subgroup of randomised (2.0%E; 95% CI: −0.2, 4.3; P=0.07) and non-randomised (1.3%E: 95% CI: −0.3, 3.0; P=0.38) studies. A pooled mean decrease in protein intake of −0.8%E (95% CI: −1.5, −0.1; P=0.02) with low heterogeneity (Q=1.06, d.f.=1, P=0.30, I2=5.2%) was found for the PSD condition compared with control, as shown in Figure 4b. A sensitivity analysis showed that the pooled mean decrease was only significant in the non-randomised studies (−1.0%E; 95% CI: −2.1, 0.0; P=0.06). As shown in Figure 4c, carbohydrate intake was not significantly different between condition in the overall meta-analysis (−0.2%E; 95% CI: −1.7, 1.3; P=0.08), with low heterogeneity (Q=3.95, d.f.=6, P=0.68, I2=0%), and in the subgroup analyses investigating randomised (−0.5%E; 95% CI: −3.1, 2.1− P=0.69) and non-randomised studies (−0.0%E; 95% CI: −1.9, 1.8; P=0.98).

Figure 4
Figure 4

(a) Forest plot of the difference in fat intake as a percentage of 24-h total EI changes after PSD in control and intervention groups. (b) Forest plot of the difference in protein intake as a percentage of 24-h total EI changes after PSD in control and intervention groups. (c) Forest plot of the difference in carbohydrate intake as a percentage of 24-h total EI changes after PSD in control and intervention groups.

Effect of PSD on RMR

Qualitative analysis

Four studies including 65 participants assessed the effects of PSD on RMR.23, 24, 29, 32 In all the studies, RMR was measured by a form of indirect calorimetry, including ventilated hood,23, 24 spirometry29 or a whole-room indirect calorimeter.32 No significant difference in RMR between conditions was detected.23, 24, 29, 32

Quantitative analysis

Four studies including n=61 in the PSD and n=61 in the control groups provided sufficient data for inclusion in the meta-analysis investigating the effect of PSD on RMR.23, 24, 29, 32 The pooled mean difference in RMR between the PSD and control conditions was statistically nonsignificant (−8 kcal per day; 95% CI: −62, 46; P=0.77), with low heterogeneity (Q=1.2, d.f.=3, P=0.75, I2=0%) (Figure 5). This lack of significant effect remained in a subgroup analysis including only the randomised studies24, 29, 32 (−16 kcal per day; 95% CI: −79, 48; P=0.63).

Figure 5
Figure 5

Forest plot of the difference in RMR (kcal per day) changes after PSD in control and intervention groups.

Quality assessment and risk of bias

Of the randomised studies,24, 25, 26, 27, 29, 32, 33 only one disclosed the method of random sequence generation and were thus at low risk of selection bias26 (Figure 6). All studies were at unclear risk of selection bias as allocation concealment was unreported. As these are sleep intervention studies, it is impossible to blind participants to treatment, which may influence performance bias. However, two studies reported that participants were unaware of the outcomes of interest.25, 30 Only one study specified that participants were aware of their assigned sleep schedule before each condition, placing it at high risk of bias.26 One study reported that outcome assessors were blinded to treatment;27 all other studies were at unclear risk of detection bias. All studies reported outcome data, and were at low risk of attrition bias; however, the largest study (n=225) only assessed outcomes in a subset (n=31 PSD, n=6 control) and did not mention drop-out rates for the remaining recruited subjects, and was thus at unclear risk of attrition bias.33 As original protocols were not available for all studies, it is difficult to assess if outcomes were selectively reported. Visual inspection of Begg’s funnel plots showed that publication bias was unlikely; however, the small number of studies included in the review precludes ascertaining publication bias.

Figure 6
Figure 6

Review author’s judgement on each risk of bias item from the Cochrane Tool for each included study in the systematic review and meta-analysis assessing the effect of PSD on energy balance.

Discussion

Our findings from this systematic review and meta-analysis of intervention studies suggest that PSD may lead to a net positive energy balance of 385 kcal per day because of a significant increase in total 24-h EI, and no effect on total 24-h EE. The perceived positive energy balance may therefore contribute to the occurrence of weight gain in those with short sleep duration. Our findings also suggest possible modest shifts in macronutrient distribution, favouring fat intake at the cost of protein. However, our results are mainly based on studies with highly restrictive sleep schedules conducted in controlled laboratory conditions over a short period of time (1 day to 2 weeks). It remains unknown whether the observed net positive energy balance is evident over a prolonged period of less restrictive sleep deprivation that mirrors the effects of chronic sleep debt. In support of our findings, a recent systematic review by Capers et al.17 who studied the impact of sleep duration on adiposity and components of energy balance concluded that sleep deprivation appears to increase food intake. Unlike the present review, Capers et al.17 did not perform a meta-analysis for EI owing to the inconsistencies of methodology to measure food intake in the six RCTs they identified. However, our meta-analysis suggests the variation in methods to measure EI did not introduce heterogeneity. Unlike the present study, Capers et al.17 found no significant effects of sleep restriction on total EE in their meta-analysis. However, in contrast to the present review, Capers et al.17 included only RCTs, considering all forms of manipulation of sleep duration including total sleep deprivation and sleep extension, and did not exclude studies in children or adolescents, which could explain the differences in results obtained.

It has been previously suggested that sleep deprivation may cause dysregulation of the hormones leptin and ghrelin, leading to an increase in EI.17, 23, 52, 53, 54 Arguably, a more plausible explanation for the observed increase in EI after PSD in this review is that this is hedonically driven. Hedonic influence was demonstrated in a crossover RCT in 26 healthy adults in which PSD resulted in greater neuronal activation in response to food stimuli, particularly in areas associated with reward.55 These findings suggest that short sleep heighten the motivation to seek food for reward.55 Furthermore, a cross-sectional study in 115 healthy, premenopausal women associated sensitivity to reward with a greater preference for foods high in fat and sugar.56 Indeed, our meta-analysis detects this possible heightened inclination for intake of fat. Yet, the longest study in the present review (2 weeks) found an increased consumption of snacks high in carbohydrates during the PSD condition with no effect on 24-h EI, yet whether the carbohydrate increase was specifically attributed to sugar or refined carbohydrates was unspecified.29 Only one study analysed saturated fat intake separately;24 this is of importance as in the long term the type of fat consumed may have serious ramifications on cardiovascular health.57, 58, 59 Laboratory-controlled conditions are necessary and valuable; however, free-living conditions may also provide information that is arguably more representative of dietary behaviours because of unlimited food choice and variety; two of the four studies including free-living conditions in this review report no significant change in 24-h EI as a result of PSD,30, 48 of which one reports a significant increase in fat intake.30 Moreover, investigating the chronic effects of PSD that is representative of habitual trends is warranted, as the adaptive response of sleep debt build-up may be to reshift toward energy balance, with long-term effects on macronutrient distribution.

We found no significant change in total 24-h EE; it is possible that this is because of differences in methods of EE measurements across studies. The two studies that used whole-room indirect calorimetry28, 32 found a similar 5% increase in 24-h EE, and this is in agreement with previous studies conducted in adolescents.60 Shechter et al.32 discerned the increase was only evident during nocturnal hours spent awake. Previous findings have shown that the metabolic costs of sleep have been shown to be lower compared with those of awake time.61, 62, 63 Despite the benefit of such accurate measuring techniques, respiratory chambers limit the ability to depict habitual physical activity owing to confined living, which is of crucial importance as activity EE is a key component of total EE. Free-living conditions may provide insight into the effect of PSD on physical activity and activity EE, and would illustrate ecologically valid integrated effects of EI and EE. For example, although Schmid et al.30 did not report on EE, they have shown that PSD resulted in decreased physical activity and intensity, measured by accelerometry. The effects of PSD on activity EE and physical activity are of importance as this may be another potential factor linking short sleep and metabolic consequences. The lack of impact of PSD on RMR observed in the present review should be interpreted with caution; there was a small number of studies identified thereby limiting study power. Moreover, the longest study assessing RMR was 2 weeks long per condition,29 and it is possible that this time frame was insufficient to allow for adaptation of the metabolic rate. Previous observational evidence reported that measured RMR was significantly elevated in chronic insomniacs compared with age-, sex- and weight-matched normal sleepers at all measured time-points over a 36-h sleep laboratory stay,64 warranting further investigation of RMR after extended periods of PSD.

Strengths and limitations

This review has some notable strengths; it comprehensively and systematically identified relevant studies, with no search restrictions and was conducted in line with the PRISMA statement. Moreover, two independent reviewers reviewed the titles and abstracts identified. A methodological limitation of this review is that not all identified studies could be included in the meta-analyses; this was a result of the reported data being inextractible and unavailable when authors were contacted. Because the majority of PSD studies that observed no effect on total 24-h EI, our pooled results could possibly overestimate EI. Our limitation highlights the need for future studies to report accurately findings to allow pooling of the data. Moreover, studies reported the absolute change in kcal and this may be dependent on an individual, yet the percentage change for EI and EE may not, emphasising the importance for future studies to also report percentage changes.

There are also various strengths and limitations related to the studies included in this review. Many of the included studies were of a robust randomised crossover design. However, there was heterogeneity in the interventions, including intervention duration, degree of sleep restriction, sleep schedule and time in bed in the control condition. Moreover, EI and EE assessment methods may have caused discrepancies in results; it is possible that laboratory-setting studies introduce bias because of the defined food choices available, and that detection of changes in total EE was limited by measurement device accuracy. In addition, it has been previously shown that the female menstrual cycle phase may impact sleep65 and components of energy balance;66, 67, 68, 69 however, only two studies in this review attempted to account for this variation in females participants.28, 31 The quality of studies identified is also a limitation to be considered; the majority of the studies were at an unclear risk of selection, performance and detection bias, mainly due to non-disclosure of method of random assignment, and lack of blinding of assessors.

Last, the external validity of included studies could be questioned as most studies were performed under experimental sessions, and included subjects were predominantly of Caucasian or African-American ethnicity. It should also be noted that the results of the present review cannot be extended to claim that extension of sleep would cause a reverse shift in energy balance, and this remains to be explored.

Conclusion

The evidence in the present review suggests that PSD may result in an increased EI, leading to a net positive energy balance of 385 kcal per day. In the long term, this may implicate weight gain; however, it remains to be investigated. We found no significant change in EE, although this may be attributed to variations in measurement techniques across studies. Short-term PSD resulted in no change in RMR, although various studies report that extended wake time is more energy costly in controlled laboratory conditions. Further investigation of the effects of PSD on physical activity and activity EE is warranted. Robustly designed RCTs are necessary to identify whether mild, chronic PSD also causes shifts in dietary intake and macronutrient distribution, with a particular focus on saturated fat and added sugar, as the evidence is scarce. Our study highlights the evident lack of studies investigating effects of PSD on energy balance in an ecologically translatable setting, particularly as free-living conditions are crucial to underpin effects on activity levels and food choice. The pragmatic way forward may be to examine whether sleep extension in habitually short sleepers can mitigate the observed effects on energy balance. Our results propose that sleep may be a potential novel target for weight management in addition to physical activity and dietary management in the clinical setting.

References

  1. 1.

    , , , , , et al. National Sleep Foundation’s sleep time duration recommendations: methodology and results summary. Sleep Health 2015; 1: 40–43.

  2. 2.

    , , . In search of lost sleep: Secular trends in the sleep time of school-aged children and adolescents. Sleep Med Rev 2012; 16: 203–211.

  3. 3.

    , , . Secular trends in adult sleep duration: a systematic review. Sleep Med Rev 2012; 16: 223–230.

  4. 4.

    , , . Sleeping at the limits: The changing prevalence of short and long sleep durations in 10 countries. Am J Epidemiol 2013; 177: 826–833.

  5. 5.

    , . Impacts of shift work on sleep and circadian rhythms. Pathol Biol 2014; 62: 292–301.

  6. 6.

    , . Sleep duration in the united states: a cross-sectional population-based study. Am J Epidemiol 2009; 169: 1052–1063.

  7. 7.

    . Impact of lifestyle and technology developments on sleep. Nat Sci Sleep 2012; 4: 19–31.

  8. 8.

    , , , , , et al. Association of sleep duration with mortality from cardiovascular disease and other causes for Japanese men and women: the JACC study. Sleep 2009; 32: 295–301.

  9. 9.

    , , , , . Short sleep is associated with obesity among truck drivers. Chronobiol Int 2006; 23: 1295–1303.

  10. 10.

    , , , , , et al. Meta-analysis of short sleep duration and obesity in children and adults. Sleep 2008; 31: 619–626.

  11. 11.

    , . Short sleep duration and weight gain: a systematic review. Obesity 2008; 16: 643–653.

  12. 12.

    , , , . Quantity and quality of sleep and incidence of type 2 diabetes a systematic review and meta-analysis. Diabetes Care 2010; 33: 414–420.

  13. 13.

    , , , , . Sleep duration predicts cardiovascular outcomes: a systematic review and meta-analysis of prospective studies. Eur Heart J 2011; 32: 1484–1492.

  14. 14.

    , , , , . Short sleep duration is associated with hypertension risk among adults: a systematic review and meta-analysis. Hypertens Res 2012; 35: 1012–1018.

  15. 15.

    , . Longitudinal associations between sleep duration and subsequent weight gain: a systematic review. Sleep Med Rev 2012; 16: 231–241.

  16. 16.

    , , . Is sleep duration related to obesity? A critical review of the epidemiological evidence. Sleep Med Rev 2008; 12: 289–298.

  17. 17.

    , , , , . A systemic review and meta-analysis of randomized controlled trials of the impact of sleep duration on adiposity and components of energy balance. Obes Rev 2015; 16: 771–782.

  18. 18.

    , . Cochrane Handbook for Systematic Reviews of Interventions Version 5.1. The Cochrane Collaboration, 2011, Available from .

  19. 19.

    , , , , , et al. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions. Ann Intern Med 2009; 15: W65–W94.

  20. 20.

    , . Chapter 8: Assessing risk of bias in included studies. Cochrane Handbook for Systematic Reviews of InterventionsVersion 5.1.0 (updated March 2011). The Cochrane Collaboration2011, Available from .

  21. 21.

    , . Quantifying heterogeneity in a meta-analysis. Stat Med 2002; 21: 1539–1558.

  22. 22.

    , , , . Bias in meta-analysis detected by a simple, graphical test. BMJ 1997; 315: 629–634.

  23. 23.

    , , , , , et al. Influence of partial sleep deprivation on energy balance and insulin sensitivity in healthy women. Obes Facts 2008; 1: 266–273.

  24. 24.

    , , , , , et al. Short sleep duration increases energy intakes but does not change energy expenditure in normal-weight individuals. Am J Clin Nutr 2011; 94: 410–416.

  25. 25.

    , , , , . Acute partial sleep deprivation increases food intake in healthy men. Am J Clin Nutr 2010; 91: 1550–1559.

  26. 26.

    , , , , , et al. Effects of experimental sleep restriction on caloric intake and activity energy expenditure. Chest 2013; 144: 79–86.

  27. 27.

    , , , , , et al. Acute changes in sleep duration on eating behaviors and appetite-regulating hormones in overweight/obese adults. Behav Sleep Med 2014; 13: 424–436.

  28. 28.

    , , , , , et al. Impact of insufficient sleep on total daily energy expenditure, food intake, and weight gain. Proc Natl Acad Sci USA 2013; 110: 5695–5700.

  29. 29.

    , , , , , . Sleep curtailment is accompanied by increased intake of calories from snacks. Am J Clin Nutr 2009; 89: 126–133.

  30. 30.

    , , , , , et al. Short-term sleep loss decreases physical activity under free-living conditions but does not increase food intake under time-deprived laboratory conditions in healthy men. Am J Clin Nutr 2009; 90: 1476–1482.

  31. 31.

    , , , , . Postprandial thermogenesis and substrate oxidation are unaffected by sleep restriction. Int J Obes 2013; 38: 1153–1158.

  32. 32.

    , , , . Experimental sleep curtailment causes wake-dependent increases in 24-h energy expenditure as measured by whole-room indirect calorimetry. Am J Clin Nutr 2013; 98: 1433–1439.

  33. 33.

    , , . Effects of experimental sleep restriction on weight gain, caloric intake, and meal timing in healthy adults. Sleep 2013; 36: 981–990.

  34. 34.

    , , . Sex and race differences in caloric intake during sleep restriction in healthy adults. Am J Clin Nutr 2014; 100: 559–566.

  35. 35.

    , , , , . Acute partial sleep deprivation increases food intake in healthy young men. Fund Clin Pharm 2011; 25: 26 (abstract).

  36. 36.

    , , , . Insufficient sleep increases caloric intake but not energy expenditure. Circulation 2012; 125: S1 (abstract).

  37. 37.

    , , , , , . Effect of sleep restriction on whole body energy expenditure in humans. Sleep 2011; 34: A41 (abstract).

  38. 38.

    , , . Short sleep duration tends to lower resting metabolic rate and peak activity level relative to habitual sleep in normal weight subjects. Obesity 2010; 18: S83 (abstract).

  39. 39.

    , , , , , . Altered nocturnal sleep architecture in response to partial sleep deprivation is associated with increased carbohydrate intake. Sleep 2012; 35: A104 (abstract).

  40. 40.

    , , . Sleep restriction associates with increased caloric intake and delayed meal timing in healthy adults. Sleep 2013; 36: A106 (abstract).

  41. 41.

    , , . Baseline slow-wave sleep negatively relates to energy balance responses during leep restriction in healthy adults. Sleep 2014; 37: A50 (abstract).

  42. 42.

    , , . Sleep restriction associates with increased food intake, weight gain and changes in food cravings in healthy adults. Sleep 2012; 35: A105 (abstract).

  43. 43.

    , , . Effects of sleep restriction on body weight and food intake in healthy adults. Appetite 2012; 59: e51 (abstract).

  44. 44.

    , , . Effects of two five-day bouts of chronic sleep restriction on caloric intake in healthy adults. Sleep 2013; 36: A87 (abstract).

  45. 45.

    , , , . Stability of energy balance responses to sleep restriction over long time intervals. Sleep 2014; 37: A47 (abstract).

  46. 46.

    , , , , , , , . Effects of acute changes in scheduled sleep duration on eating behavior. Obesity 2011; 19: S194 (abstract).

  47. 47.

    , , , , . Effect of sleep restriction on physical functions—a respiratory chamber study. Sleep 2013; 36: A105 (abstract).

  48. 48.

    , , , , , . The relationship between sleep deprivation and the energy balance pathways of diet and physical activity. Med Sci Sports Exerc 2013; 45: 336 (abstract).

  49. 49.

    , , , , . Sleep curtailement in healthy young adults is associated with increased ad lib food intake. Sleep 2009; 32: A131 (abstract).

  50. 50.

    , , , , , , Dietary intake of sleep-deprived, on-call anesthesiology residents. FASEB J 2013; 27: 1064–1 (abstract).

  51. 51.

    , , , , . Sleep restriction decreases the physical activity of adults at risk for type 2 diabetes. Sleep 2012; 35: 977–984.

  52. 52.

    . The role of sleep duration in the regulation of energy balance: effects on energy intakes and expenditure. J Clin Sleep Med 2013; 9: 73–80.

  53. 53.

    , , , , , et al. History of the development of sleep medicine in the United States. J Clin Sleep Med 2005; 1: 61–82.

  54. 54.

    , , , . Article Brief Communication: sleep curtailment in healthy young men is associated with decreased leptin levels, elevated ghrelin levels, and increased hunger and appetite. Ann Intern Med 2004; 141: 846–851.

  55. 55.

    , , , , , . Sleep restriction leads to increased activation of brain regions sensitive. Am J Clin Nutr 2012; 95: 818–824.

  56. 56.

    , , , , , . From motivation to behaviour: a model of reward sensitivity, overeating, and food preferences in the risk profile for obesity. Appetite 2007; 48: 12–19.

  57. 57.

    , , . Effects on coronary heart disease of increasing polyunsaturated fat in place of saturated fat: a systematic review and meta-analysis of randomized controlled trials. PLoS Med 2010; 7: 1–10.

  58. 58.

    , , , , , et al. Major types of dietary fat and risk of coronary heart disease: a pooled analysis of 11 cohort studies. Am J Clin Nutr 2009; 89: 1425–1433.

  59. 59.

    , , . Effects of monounsaturated fatty acids on cardiovascular risk factors: a systematic review and meta-analysis. Ann Nutr Metab 2011; 59: 176–186.

  60. 60.

    , , , , , . Sleep restriction is not associated with a positive energy balance in adolescent boys. Am J Clin Nutr 2012; 96: 240–248.

  61. 61.

    , , , . Energy expenditure durign sleep in men and women: evaporative and sensible heat losses. Hum Nutr Clin Nutr 1987; 41: 225–233.

  62. 62.

    , , , . Overnight and basal metabolic rates in men and women. Eur J Clin Nutr 1988; 42: 137–144.

  63. 63.

    , . Relationship between overnight energy expenditure and BMR measured in a room-sized calorimeter. Eur J Clin Nutr 1999; 53: 107–111.

  64. 64.

    , . 24-Hour metabolic rate in insomniacs and matched normal sleepers. Sleep. Sleep NY 1995; 18: 581–588.

  65. 65.

    , , . Circadian variation of sleep during the follicular and luteal phases of the menstrual cycle. Sleep 2010; 33: 647–656.

  66. 66.

    . The effect of the menstrual cycle on patterns of food intake. Am J Clin Nutr 1981; 34: 1811–1815.

  67. 67.

    . 24-Hour energy expenditure menstrual. Am J Clin Nutr 1986; 44: 614–619.

  68. 68.

    , , . Changes in energy expenditure during the menstrual cycle. Br J Nutr 1989; 61: 187–199.

  69. 69.

    , , , , , . Resting metabolic rate and thermic effect of a meal in the follicular and luteal phases of the menstrual cycle in well-nourished Indian women. Am J Clin Nutr 1995; 61: 296–302.

  70. 70.

    , , , . Insufficient sleep increases caloric intake but not energy expenditure. Circulation 2012; 1 (abstract).

  71. 71.

    , , , , , . Partial sleep deprivation and energy balance in adults: an emerging issue for consideration by dietetics practitioners. J Acad Nutr Diet 2012; 112: 1785–1797.

Download references

Acknowledgements

The authors’ responsibilities were as follows: GP and JD designed the study, HA, SH, JD, and GP performed the literature search and the meta-analysis. GP and JD had primary responsibility for final content. All authors were substantially involved in the writing process. All authors read and approved the final manuscript.

Author information

Author notes

    • J Darzi
    •  & G K Pot

    These authors contributed equally to this work.

Affiliations

  1. Diabetes and Nutritional Sciences Division, School of Life Sciences and Medicine, King’s College London, London, UK

    • H K Al Khatib
    • , S V Harding
    • , J Darzi
    •  & G K Pot
  2. VU University Amsterdam, Health and Life, Faculty of Earth and Life Sciences, Amsterdam, Netherlands

    • G K Pot

Authors

  1. Search for H K Al Khatib in:

  2. Search for S V Harding in:

  3. Search for J Darzi in:

  4. Search for G K Pot in:

Competing interests

The authors declare no conflict of interest.

Corresponding author

Correspondence to G K Pot.

Supplementary information

About this article

Publication history

Received

Revised

Accepted

Published

DOI

https://doi.org/10.1038/ejcn.2016.201

Supplementary Information accompanies this paper on European Journal of Clinical Nutrition website (http://www.nature.com/ejcn)

Further reading