Feasibility of sleep extension and its effect on cardiometabolic parameters in free-living settings: a systematic review and meta-analysis of experimental studies

Abstract

Aims

Inadequate sleep is a global health issue and has been associated with an increased risk for cardiovascular diseases. As a part of sleep hygiene, intentional lengthening of night-time sleep duration (i.e. sleep extension) might be a behavioural intervention to improve cardiometabolic health. To examine the feasibility of sleep extension and its effects on cardiometabolic parameters in free-living settings.

Methods and results

This review was registered in PROSPERO (CRD42019146174). Five databases were searched. Only experimental studies conducted in adults without a diagnosis of sleep disorder were included. The pooled mean difference was calculated by the inverse variance method. Narrative summaries were also used. Thirteen studies from 11 trials were included. The intervention ranged from 3 days to 6 weeks. Sleep extension increased total sleep time by 51 min [95% confidence interval (CI) 39–63]. Overall, sleep extension did not result in significant changes in blood pressure. However, sub-group analysis revealed that when 24 h mean blood pressure was obtained among those with pre-hypertension or Stage 1 hypertension, sleep extension reduced systolic (weighted mean difference = −7.8 mm/Hg; 95% CI −10.6 to −4.9), and diastolic blood pressure (weighted mean difference = −4.2 mm/Hg; 95% CI −6.7 to −1.8). The pooled effects on fasting glucose and insulin resistance were not significant. The effect of sleep extension on other parameters (e.g. heart rate) was not consistent.

Conclusion

Sleep extension is feasible and could increase sleep in free-living settings. Sleep extension shows promise for reducing 24 h mean blood pressure among those with pre-hypertension or hypertension. More large-scale studies are needed to examine its long-term effects.

Introduction

Sleep is a physiological and behavioural process, critical for good health. Sleep duration is one of the key dimensions of sleep health.¹ The National Sleep Foundation (NSF)² has recommended that adults should obtain at least 7 h of sleep. Using this recommendation, short sleep was defined as having less than 7 h of sleep in this review. Short sleep is becoming a worldwide health problem and could be caused by various factors, including unhealthy lifestyle habits (e.g. using electronic devices before sleep), environmental factors, and sleep disorders (e.g. insomnia, sleep apnoea, and irregular sleep-wake rhythm disorder).³ Depending on the methods of assessment, the prevalence of short sleep has ranged from 32%⁴ to around 65%.⁵

Coupled with the high prevalence of short sleep is the increase in the prevalence of cardiometabolic diseases. Worldwide, hypertension and diabetes remain two huge health problems, with 1 in 4–5 adults having hypertension⁶ and 1 in 11 adults living with diabetes.⁷ The prevalence of obesity and overweight was as high as 53.1%⁸ and around 25% of the adults had metabolic syndromes.⁹ Emerging evidence suggests that short sleep may increase the risk of cardiometabolic diseases such as hypertension, heart diseases, obesity, and diabetes,^3,10–12 via various pathophysiological pathways. Short sleep has been shown to alter autonomic activity, inflammation, oxidative stress, and appetite-regulation hormones, contributing to increased blood pressure, endothelial dysfunction, and glucose dysregulation.^3,13–15 Those changes, if persistent, can lead to an increased risk of cardiometabolic diseases.³

Given the adverse effects of short sleep on cardiometabolic health, questions arise about the potential benefits of intentional lengthening of night-time sleep duration (i.e. sleep extension). Based on the two-process model of sleep regulation, sleep is regulated by a circadian process and a homeostatic process. As an indicator of sleep need, slow-wave sleep is thought to reflect the homeostatic process.¹⁶ Among those with short sleep, there might be a gradual, accrual of the unsatisfied physiological need for sleep over time.¹⁷ Sleep extension may thus satisfy the unmet sleep need. According to recent studies,^18,19 a 7-day sleep extension (10 h time in bed) among healthy adults decreased the percentage of slow-wave sleep, suggesting that short-term sleep extension could decrease sleep need. Previous evidence also suggests a cross-talk between sleep homeostasis and metabolic cascades.²⁰ Thus, it is likely that sleep extension could preserve cardiometabolic health by regaining sleep homeostasis.

Empirical studies have been conducted to investigate the effect of sleep extension. For instance, among adults who are habitually short sleepers, a 4-week sleep extension in the free-living setting was found feasible and changed the participants’ eating behaviours.²¹ In another group of healthy adults, an increase in fasting insulin sensitivity was observed after a 6-week sleep extension at home.²² With the number of sleep extension studies increasing, reviews have been conducted to synthesize the findings. Specifically, Pizinger et al.²³ reviewed five studies and indicated that sleep extension may help to reduce the risk for chronic cardiometabolic disorders. Recently, Henst et al.²⁴ conducted a systematic review of seven trials examining the impact of sleep extension. It was suggested that sleep extension may improve insulin sensitivity, appetite, and dietary intake.

Sleep extension has been delivered in both habitual short sleepers and those with normal sleep duration. These individuals may have different needs and abilities to obtain extra sleep, particularly in the free-living setting where the participants were unsupervised (e.g. at home). While the aforementioned reviews provided preliminary evidence for the role of sleep extension on cardiometabolic health, there remains a need to synthesize current evidence while taking that into consideration. Additionally, the effect size of sleep extension remains unknown and a meta-analysis enables quantification of the effect size across studies.²⁵ Thus, the aim of this systematic review and meta-analysis was: (i) to quantify the effect of sleep extension on improving sleep duration in the free-living setting among both habitual short sleepers and those with normal sleep duration; (ii) to investigate the effect of sleep extension on various cardiometabolic outcomes (e.g. blood pressure and glucose metabolism). Inadequate sleep is becoming a global health issue and sleep extension is a lifestyle intervention that could be delivered by various healthcare professionals including nurses. Findings from this review may provide evidence for future research and nursing practice.

Methods

We developed a review protocol following the International Prospective Register of Systematic Reviews PROSPERO guidelines and registered it online (CRD42019146174). This review was reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.²⁶

Search strategy

The search strategy (e.g. databases and search terms) was developed by the team based on knowledge in the field and prior systematic reviews on the same topic. One author (B.Z.) performed a systematic search in electronic databases from inception to July 2019. The databases included PubMed, CINAHL, Web of Science, Embase, and Cochrane. These databases cover biomedical research, life sciences, evidence-based research, nursing and allied health, and many different academic disciplines. They also cover grey literature such as conference abstracts, dissertations/theses, and ongoing (or completed but not unpublished) trials. No restriction on the year of publication was applied. The search was restricted to the English language. The following search term was used: ‘sleep extension’ OR ‘extended sleep’ OR ‘sleep manipulation’ OR ‘sleep intervention’ OR ‘time in bed extension’ OR ‘bed time extension’. The search term was applied to Title/Abstract [PubMed], Abstract [CINAHL], Topic [Web of Science], and Title/Abstract/Keyword [Embase and Cochrane]. Identified records were managed using Endnote and duplicates were removed by the ‘finding duplicates’ function in endnote and manual removal. The second round of systematic search was conducted by one author (B.Z.) to identify studies published between 2019 and June 2020.

Screening process

We included studies conducted in healthy adults who were ≥18 years. The following exclusion criteria were applied: (i) studies that included participants with known sleep disorders (e.g. insomnia, sleep apnoea, or narcolepsy) or were shift workers, (ii) studies that used other interventions including pharmacological treatment, cognitive behavioural therapy aiming at improving sleep quality, recovery sleep after sleep restriction or deprivation, and mixed interventions where the effect of sleep extension cannot be separated from the other components, (iii) studies that did not assess any of the outcomes of interest as presented in the ‘primary and secondary outcomes’ section below, (iv) studies conducted in the laboratory setting, and (v) review papers, qualitative studies, conference abstracts, editorials, or case reports.

We selected the studies following the PRISMA flowchart.²⁶ Initial screening was performed by two independent reviewers (B.Z. and C.S.) based on the title/abstract. The full texts of the potentially eligible studies were retrieved for further screening. Two independent reviewers (B.Z. and C.S.) read the full texts and determined the final inclusion based on the inclusion and exclusion criteria listed above. The decision process was recorded using pre-determined rules developed combining the features in EndNote (e.g. one-star rating means excluded for inappropriate populations; five-star rating means excluded for outcomes of interest not reported). We kept a paper trail to record this process and any discrepancy between the two reviewers was resolved by discussion. During screening the title/abstract, a discrepancy occurred between the two reviewers in the eligibility of eight studies. After discussion, they decided to include all of them for full-text retrieval. During screening the full-text, discrepancy occurred in whether two studies in people with pre-hypertension or stage 1 hypertension should be included. The discrepancy was not resolved by discussion. Thus, a third reviewer (J.C.) was consulted to settle the disagreement. Bibliographies of included studies and prior similar systematic reviews (other sources) were hand searched for additional eligible studies.

Data extraction and presentation

Two independent reviewers (B.Z. and C.S.) extracted the key information. The extracted data included: (i) study characteristics (e.g. first author, year, country, design, and sample size), (ii) participant characteristics (e.g. age, sex, BMI), (iii) interventions, (iv) outcomes (measures), (v) feasibility data including participant compliance and effect on total sleep time (TST), and (vi) main findings. The corresponding authors were contacted if the information we needed for the analysis was not reported in the paper. Data displayed in graphs were extracted using WebPlotDigitizer (V4.1)²⁷ if they cannot be obtained by contacting the authors. This method has demonstrated high levels of validity, reliability, and usability. ²⁸ Discrepancies were resolved by a third reviewer (J.C.).

A data matrix was developed to organize and present the extracted information. To clarify the effect of sleep extension on improving TST in habitual short sleepers and those with normal sleep duration, we stratified and tabulated the information based on whether the participants were habitual short sleepers. Currently, there has been no consensus on the definition of short sleep. Based on the NSF recommendation² and sleep term classification proposed by Grandner et al.,²⁹ we defined short sleep as chronic, habitual sleep duration of 7 h or less.

Primary and secondary outcomes

The primary outcomes were cardiometabolic parameters including insulin resistance or sensitivity, fasting glucose, blood pressure, and heart rate. The secondary outcomes included energy balance, subjective appetite or hunger, appetite- and stress-related hormones, inflammatory biomarkers, microbiome, and weight. We also assessed the effect of the intervention on increasing TST.

Quality appraisal

Two independent reviewers (B.Z. and C.S.) performed quality appraisals using the Cochrane Risk of Bias Tool.³⁰ The following five areas were assessed: (i) selection, (i) performance, (iii) detection, (iv) attrition, and (v) reporting bias. The quality appraisal process was recorded using a Word document. Prior to performing the appraisal, two reviewers (B.Z. and C.S.) created two working tables. The first one was used to document the detailed decision-making process including scores for each area and corresponding rationales. The second one was used to present the scores for each study. Upon completion, scores rated by the two reviewers were compared. The discrepancy was resolved by discussion and further disagreement was resolved by a third reviewer (I.B.B.).

Data syntheses

Stata 13.0 (StataCorp LP, College Station, TX, USA) was used for statistical analyses. Mean and standard deviation (SD) from each study were used for analyses. If not reported, SD was calculated from the standard error of the mean or 95% confidence interval (95% CI). The post-intervention values or changes from baseline were used wherever applicable. The sum score from cross-over randomized clinical trials (RCTs) was used as if the trial used a parallel-group design. This conservative approach may under-weight instead of over-weight the studies.³¹ The pooled mean difference (MD) with 95% CI was calculated by the inverse variance weighted average method for outcomes reported in three or more studies. A forest plot was used to present the results. Weighted mean difference (WMD) was calculated when the outcome was measured using the same unit. Otherwise, standardized mean difference (SMD) was calculated by Hedge’s g method to adjust for the overestimation of the mean difference in small trials.³² The selection between the fixed- and random-effect models was determined based on a priori heterogeneity test as well as the nature of the studies (e.g. heterogeneity in the study population). The degree of heterogeneity was examined by I² value.³³ A random-effect model was used when severe heterogeneity was detected, and a fixed-effect model was used otherwise. When severe heterogeneity was present, potential sources of heterogeneity were explored by removing possible outlying studies. Sub-group analyses were performed wherever applicable. Statistical significance was set at P < 0.05. We used qualitative narratives if data could not be pooled.

Results

The literature search process is shown in Figure 1. The initial search yielded 1174 records and the second round of search resulted in 159 records. A total of 65 studies were retrieved for full-text review, among which 52 were excluded for reasons listed in Figure 1. Thus, 13 studies (n = 259) from 11 trials were included in the final analysis.

Study and participant characteristics

Table 1 shows the characteristics of the studies and participants. The studies were published between 2011 and 2020 in the USA,^{34,35,39,42,43} UK,²¹ Belgium,²² Thailand,^37,38 Japan,³⁶ France,^40,41 and New Zealand.⁴⁴ So-ngern et al.³⁷ and Reutrakul et al.³⁸ studies were from the same trial. Arnal et al.⁴⁰ and Chennaoui et al.⁴¹ studies were from the same trial. Various design was used including parallel-group RCT (n = 4), cross-over RCT (n = 4), and non-RCT (n = 5). The sample size was between 8 and 53.

The participants did not have a diagnosis of sleep disorders (e.g. insomnia and sleep apnoea). They were screened for extreme chronotype or irregular sleep patterns by sleep diary, actigraphy, or questionnaire. Two trials^40,44 included males only and the remaining included both males and females. Two studies included adults with pre-hypertension or Stage 1 hypertension.^34,35 The mean age of the participants was between 20.5 years and 48.4 years. The mean BMI was between 21.8 and 28.0 kg/m². For studies that included only habitual short sleepers, several thresholds were used, including less than 6 h, 6.5 h, or 7 h sleep per night (Table 1).

Quality appraisal

The quality appraisal outcomes are shown in Table 2. Briefly, for studies that used a pre-post design, a high risk of bias on randomization was scored. It was not feasible to blind the participants. Thus, the performance bias was rated high for all studies. However, strategies were used to minimize the performance bias as shown in Table 2. In terms of detection bias, assessor blinding was not used, but only objective outcomes were obtained in several studies and a low risk of performance bias was thus assigned to these studies. A high risk of attrition bias was observed in two studies.^42,44

Characteristics of the intervention

The intervention protocol varied across studies (Table 1). The intervention was developed by researchers from various disciplines including psychology, neuroscience, sleep medicine, nutritional science, nursing, endocrinology, occupational health, sports science, and behavioural science. The duration of the sleep extension ranged from 3 days³⁶ to 6 weeks.^22,34,35 The participants were usually provided with sleep extension alongside sleep hygiene strategies. Mostly, they were instructed to extend their time in bed by 1 h.^{21,22,34,35,37,43} Some were instructed to extend time in bed by 3 h⁴² and others were instructed to extend time in bed to 8 h,³⁶ 8.5 h,³⁹ or 10 h^40,44 per night. Four studies^21,22,37,40 extended sleep by advancing bedtime only. The remaining studies extended sleep by individualizing both bedtime and wakeup time. Sleep was assessed by actigraphy in three studies^34,36,40 and by a combination of actigraphy and sleep diary in the remaining studies.

Outcome assessment

During the baseline and after the intervention, outcomes of interest were measured. The outcomes included blood pressure, heart rate, and indicators related to glucose-regulation, such as fasting glucose, insulin resistance, and glucose area under the curve (AUC) which is an index of whole glucose excursion after glucose loading and has been used as a measure of glucose intolerance.⁴⁵ Other metabolic-related parameters included appetite, appetite-regulating hormones, stress hormones, growth hormones, gut microbiome, and BMI or weight. Detailed measures are presented in Table 1.

Impact of sleep extension on sleep duration

Participant compliance varied across studies, with an attrition rate ranging from 0% to 33.3% (Table 3). For habitual short sleepers, overall, the intervention was effective in increasing TST. Specifically, sleep extension increased TST by approximately 35 min in two studies^34,35 and over 60 min in another two studies.^37,39 The group difference in TST was as high as 66 min²² and 2 h³⁶ in the other two studies. In comparison, in Al Khatib et al. study,²¹ sleep extension only increased TST by 21 min. For those with normal sleep duration, similarly, the intervention increased TST by 43 min⁴³ to 2 h.⁴²

To better visualize the effect of sleep extension on TST, a forest plot was used. For studies from the same trial, only one was included in the meta-analysis. One study³¹ was excluded from the analysis as its intervention was very different from the remaining. Thus, 10 studies were pooled. There remained severe heterogeneity between studies, and the random-effect model was used. Based on Figure 2, overall, sleep extension increased TST by 51 min (95% CI 39–63). For habitual short sleepers, sleep extension increased TST by 50 min (95% CI 35–64). A 54 min increase was found for those with normal sleep duration (95% CI 37–71).

Effect of sleep extension on blood pressure and heart rate

The effects of sleep extension on blood pressure and heart rate are shown in Table 3. Results from six studies^{21,34–36,42,43} were pooled for blood pressure (Figure 3). Overall, sleep extension did not result in significant changes in systolic (WMD = −2.9 mm/Hg; 95% CI −7.8 to 2.1; P = 0.26) and diastolic blood pressure (WMD = −1.6 mm/Hg; 95% CI −4.2 to 1.0; P = 0.22). However, sub-group analysis indicated that when 24 h mean of blood pressure was obtained among those with pre-hypertension or Stage 1 hypertension, sleep extension was effective in reducing systolic (Figure 3A, WMD = −7.8 mm/Hg; 95% CI −10.6 to −4.9; P < 0.001) and diastolic blood pressure (Figure 3B, WMD = −4.2 mm/Hg; 95% CI −6.7 to −1.8; P = 0.001). No significant effect of sleep extension on heart rate^35,42,43 or heart rate variability²¹ was found.

Effect of sleep extension on glucose metabolism-related parameters

Parameters related to glucose metabolism were evaluated in four studies^21,22,37,41 including fasting glucose, insulin resistance, glucose AUC, and fasting insulin. All four studies examined the impact of sleep extension on fasting glucose. One study using a pre–post design²² found that fasting glucose levels did not differ between the two sleep conditions. However, a significant association was observed between pre- and post-intervention percent change in TST and that in fasting glucose (r = 0.53, P = 0.041). In this study, no detailed data were provided. We tried to contact the authors via email but did not get a response. Thus, this study²² was not included in the meta-analysis. Based on the pooled results from three studies^21,37,41 (Figure 4A), sleep extension did not have a significant effect on fasting glucose (SMD = −0.11; 95% CI −0.55 to 0.34; P = 0.64). Three studies^21,37,41 assessed insulin resistance measured by HOMA-IR. Based on the pooled results (Figure 4B), sleep extension did not result in significant changes in insulin resistance (WMD = −0.07; 95% CI −0.47 to 0.33; P = 0.74).

Two studies examined the effect of sleep extension on fasting insulin levels but did not find a significant impact.^22,41 However, in one of these two studies, a significant association between pre- and post-intervention percent change in TST and that in fasting insulin was found (r = −0.60, P = 0.025).²² In the other study, sleep extension did not result in significant changes in glucose AUC.³⁷

Effect of sleep extension on other parameters

The effects of sleep extension on other metabolic parameters are shown in Table 3. Data were not pooled due to the heterogeneity of the assessment or the limited number of studies. Three studies^21,35,37 examined the effect of sleep extension on total caloric intake assessed by food diaries, with no significant findings. In comparison, sleep extension reduced free sugar intake and carbohydrates intake²¹ and lowered overall appetite and appetite for sweet/salty food.³⁹ One study examined the impact of extension on energy expenditure but did not find a significant effect.²¹ Sleep extension did not result in significant changes in weight or BMI.^21,22,34,37

The effect of sleep extension on stress hormones has been mixed, with two studies^21,40 reporting a non-significant impact on blood cortisol levels and the other⁴⁴ reporting a small decrease in saliva cortisol after sleep extension. Sleep extension increased IL-6 in one study⁴² but did not in the other.³⁵ Interestingly, one study⁴¹ reported that after a 6-night sleep extension, subjects had higher levels of free and total IGF-I when compared with their habitual sleep. The impacts of sleep extension on gut microbiome³⁸ and growth hormones⁴¹ were not significant.

Discussion

The aim of this systematic review was to investigate the feasibility of sleep extension and its effect on cardiometabolic parameters in free-living settings. We synthesized 13 studies from 11 clinical trials and found that sleep extension significantly improved TST among both habitual short sleepers and normal sleepers. Sleep extension demonstrated potentials for reducing 24 h mean blood pressure among those with pre-hypertension or Stage 1 hypertension; however, no significant effect on glucose metabolism was found.

In this review, we found that extending sleep in free-living settings is a feasible lifestyle intervention. The duration of the intervention varied between 3 days to 6 weeks and the sample size ranged from 8 to 53, possibly contributing to the heterogeneity between studies. Interestingly, the intervention was overall shorter for normal sleepers than habitual short sleepers. This might be partially explained by the two-process model of sleep regulation. The homeostatic process of sleep regulation depends on the duration of waking/sleep and is characterized by a saturating exponential increase during wakefulness and exponential decrease during sleep.¹⁹ With sleep extension, the accumulated unmet sleep need could be satisfied. Physiologically, the sleep need for normal sleepers was smaller than that of habitual short sleepers, and thus requiring a shorter intervention. Based on the pooled analysis, the effect of sleep extension among habitual short sleepers (WMD = 50min) was comparable to that of normal sleepers (WMD = 54min). A possible reason could be that various thresholds were used to define short sleep (e.g. 6 h, 6.5 h, or 7 h sleep) in the source studies. Although we used a threshold for short sleep proposed by sleep experts, there are normal variations between individuals and over an individual’s lifespan. These variations may affect the pooled results. Overall, the intervention improved TST by approximately 51 min, suggesting that sleep extension could be scaled up in future research and clinical practice.

The effect of sleep extension on blood pressure varied based on the assessment methods. When resting blood pressure was measured by one single snapshot, sleep extension did not demonstrate a significant impact. In comparison, when ambulatory methods were used for assessment, based on the pooled analysis of two studies, 6-week sleep extension decreased 24 h mean systolic blood pressure by 7.8 mm/Hg and diastolic blood pressure by 4.2 mm/Hg. Notably, participants that received continuous monitoring of blood pressure had pre-hypertension or stage 1 hypertension. It is possible that sleep extension is only beneficial for those with hypertension. The reduction of blood pressure observed here was comparable to that of other behavioural interventions such as physical activity.^46,47 A 9-month physical activity intervention reduced systolic blood pressure by 8.7 mmHg among patients with hypertension.⁴⁶ Similarly, another 3-month physical activity program decreased systolic and diastolic blood pressure by 7.9 and 6.2 mmHg, respectively, among sedentary older adults.⁴⁷ The above evidence suggests that as a behavioural intervention, sleep extension might benefit patients with hypertension. Nevertheless, this finding needs to be interpreted with caution. The pooled results were based on a small sample (n = 38) with a short monitoring duration. Whether the impact of the 6-week sleep extension on blood pressure could be maintained remains unknown. There might be a reversion effect of the effectiveness of this intervention over time. Thus, future studies are warranted to confirm findings from this review. Such studies should examine the long-term effect of sleep extension, using ambulatory methods among patients with hypertension.

In this review, we did not find a significant effect of sleep extension on insulin resistance. This finding is in contrast to a recent meta-analysis of RCTs,¹² which demonstrated that sleep restriction increased insulin resistance. Insulin resistance has been a hallmark of metabolic disorders.^48,49 Under sleep restriction, a threshold effect has been suggested whereby more severe sleep restriction is needed to exert a more pronounced effect on glucose metabolism. ⁵⁰ Likewise, sleep duration may be extended to a larger degree to achieve a significant effect. Indeed, in one study included in this review, So-ngern et al.³⁷ found that glucose parameters did not differ between conditions when all participants were included. However, when the analysis was restricted to those who could sleep over 6 h during the intervention, the impact was significant. Similarly, in Leproult et al. study,²² pre- and post-intervention percent changes in TST were proportionate to those in glucose parameters, suggesting that a significant increase in TST might be needed to improve glucose metabolism. It is worth mentioning that participants included in this review did not have a diagnosis of sleep disorder (e.g. insomnia and sleep apnoea). They were young to middle-aged adults who likely experienced social jetlag. Social jetlag represents misalignment between social and biological times. In modern society, social schedules (e.g. work and school) could interfere with a majority of people’s sleep, causing sleep debt on workdays.⁵¹ During the weekend, people could have ‘catch-up’ sleep to compensate for sleep debt accumulated during workdays. Based on an in-lab study among men with habitual short sleep, weekend ‘catch-up’ sleep increased insulin sensitivity.⁵² It is possible that participants in the control group recuperated by having ‘catch-up’ sleep, thus contributing to the null finding in this review.

Similarly, we did not find a significant effect of sleep extension on fasting glucose. This was partially explained by the small sample size. Interestingly, other sleep manipulations (e.g. circadian misalignment or sleep restriction) caused significant changes in insulin resistance without affecting fasting glucose.^53–55 Fasting glucose may be more resilient to sleep manipulations. Additionally, the intervention may not be long enough for adaptation. Studies in this review were mostly pilot tests. A longer intervention might be needed to observe a significant change. The settings in which sleep extension was delivered could also be an important factor. In the free-living setting, participants’ diet and exercise cannot be systemically controlled. Glucose concentrations are sensitive to diet and exercise. Different diet approaches have different impacts on fasting glucose.⁵⁶ Even a single bout of aerobic exercise could reduce fasting glucose the following morning.⁵⁷ These factors may have confounded the impact of sleep extension on glucose.

Sleep restriction significantly increased caloric intake.¹² It thus could be speculated that sleep extension may decrease caloric intake. However, no significant impact of sleep extension on energy balance and weight was found. Our null finding on weight is consistent with the one from a previous meta-analysis of sleep restriction.¹² This finding could be attributed to the short duration of the intervention. It is worth mentioning that in this review, studies were typically conducted in adults with normal BMI. For this population, maintaining weight or even increasing weight (for those at the lower end of the BMI) is one of the treatment goals. Interventions, such as sleep extension, might produce different degrees of effect on weight for those with normal BMI and those who are over-weight or obese. Future interventions should take into account the sub-group variations in BMI.

Sleep restriction could increase inflammation, disrupt sympathovagal activity and cortisol release,⁵⁸ thereby impairing cardiometabolic health. Theoretically, sleep extension might counteract the disruption caused by short sleep. However, based on current evidence, increasing sleep duration did not change these parameters. Current evidence suggests that slow-wave sleep plays an important role in the regulation of cardiometabolic and immunologic functions.^59,60 It is possible that sleep extension could improve cardiometabolic health through enhancing slow-wave sleep. Nonetheless, studies included in this review only assessed TST. Future research is needed to examine whether and how sleep architecture might influence cardiometabolic parameters under sleep extension. Change in TST may cause a change in the circadian phase. Short sleep could induce circadian misalignment, where exogenous behaviours are not aligned with the endogenous circadian rhythm.^61,62 In a previous study,⁶³ participant’s sleep was restricted by delaying bedtime and advancing rise time. Although their sleep was centred to the mid-sleep time, a shift in the circadian phase occurred, which drives more night-eating behaviours. In this review, four studies^21,22,37,40 extended sleep by advancing bedtime only, and the remaining studies extended sleep by individualizing both bedtime and rise time. Sleep extension may restore the alignment between endogenous circadian rhythm and exogenous stimuli. This restoration might have beneficial effects on cardiometabolic health. Future sleep extension studies should consider the potential impact of changes in the circadian phase on outcomes of interest.

This review has strengths and limitations. Including only experimental studies has enhanced the causality inference. Sleep extension was delivered in free-living settings, which enhanced the external validity. We also quantified the impact of sleep extension on TST and main cardiometabolic indicators. This may provide information for future clinical trials. There are several limitations. Although we did an exhaustive search, the number of eligible studies remained small. That may have precluded us from detecting important relationships of interest as well as conducting sub-group analyses. Relatedly, participants included in this review were mostly young or middle-aged adults, with a mean age of 20.5 and 48.4 years. Similarly, their mean BMI ranged from 21.8 to 28.0 kg/m², representing a relatively normal weight or over-weight sample. Thus, findings from this review may not be generalized to the older population or those with obesity. Additionally, the intervention and follow-up duration were typically short. We also cannot rule out the Hawthorne effect.⁶⁴ A previous sleep extension study reported improvements in physiological and behavioural indicators between the screening and randomization phase, suggesting the presence of the Hawthorne effect.⁶⁵

Despite the above limitations, findings from this review provide directions for future research and nursing practice. Based on the preliminary evidence, sleep extension is feasible in free-living settings. Sleep extension demonstrated short-term effects on blood pressure among those with hypertension or pre-hypertension. More studies are needed to confirm findings from this review. Ideally, the studies should be large-scale trials conducted in habitual short sleepers with existing cardiometabolic diseases (e.g. hypertension or diabetes). Studies focusing on people with obesity or older adults may help to understand whether these populations could benefit from sleep extension. In addition, long-term follow-up should be performed to evaluate whether the effects of sleep extension could maintain over time. Given that short sleep among people without an underlying sleep disorder is likely a result of social jetlag, future studies should take into account the possible confounding effect of ‘catch-up’ sleep while delivering the intervention. Meanwhile, nurses are encouraged to share sleep hygiene tips with their patients.

Conclusions

Currently, there remains a paucity of large studies investigating the effect of sleep extension on cardiometabolic health. Sleep extension could increase sleep duration among both habitual short sleepers and normal sleepers in the free-living setting. Given the increasing prevalence of short sleep and its related health consequences, it is encouraging to know that sleep extension might be used as a potential behavioural intervention to decrease burdens related to short sleep. More large-scale trials with a longer duration are needed. These studies may shed more light on the relationship between sleep and cardiometabolic health. For those with hypertension, sleep extension might be considered an integral part of disease management.

Acknowledgements

We thank the authors of the original works cited in this review for providing additional data required for the analyses.

Funding

The ‘Sailing Program’ of Science and Technology Commission of Shanghai Municipality (19YF1425300 to B.Z.); Shanghai Municipal Health and Family Planning Commission (20194Y0101 to B.Z.); National Natural Science Foundation of China (71904119 to B.Z.), and Startup Fund for Youngman Research at Shanghai Jiao Tong University (19X100040042 to B.Z.), in part. Shanghai Municipal Education Commission ‘Young Eastern Scholar’ (China) to B.Z. The funding sources have no roles in study design; in the collection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the article for publication.

Conflict of interest: none declared.

References