research papers on sleeping position

Sleep timing, sleep consistency, and health in adults: a systematic review

Affiliations.

• 1 Healthy Active Living and Obesity Research Group, Children's Hospital of Eastern Ontario Research Institute, Ottawa, ON K1H 8L1, Canada.
• 2 School of Kinesiology and Health Studies, Queen's University, Kingston, ON K7L 3N6, Canada.
• 3 Department of Kinesiology, University of Waterloo, Waterloo, ON N2L 3G1, Canada.
• 4 Department of Applied Human Sciences, University of Prince Edward Island, Charlottetown, PE C1A 4P3, Canada.
• 5 Independent Researcher, Kanata, ON K2K 0E5, Canada.
• 6 School of Rehabilitation Sciences, McMaster University, Hamilton, ON L8S 1C7, Canada.
• 7 Queen's University Library, Queen's University, Kingston, ON K7L 3N6, Canada.
• 8 Départment de psychologie, Université de Montréal, Montreal, QC H2V 2S9, Canada.
• PMID: 33054339
• DOI: 10.1139/apnm-2020-0032

The objective of this systematic review was to examine the associations between sleep timing (e.g., bedtime/wake-up time, midpoint of sleep), sleep consistency/regularity (e.g., intra-individual variability in sleep duration, social jetlag, catch-up sleep), and health outcomes in adults aged 18 years and older. Four electronic databases were searched in December 2018 for articles published in the previous 10 years. Fourteen health outcomes were examined. A total of 41 articles, including 92 340 unique participants from 14 countries, met inclusion criteria. Sleep was assessed objectively in 37% of studies and subjectively in 63% of studies. Findings suggest that later sleep timing and greater sleep variability were generally associated with adverse health outcomes. However, because most studies reported linear associations, it was not possible to identify thresholds for "late sleep timing" or "large sleep variability". In addition, social jetlag was associated with adverse health outcomes, while weekend catch-up sleep was associated with better health outcomes. The quality of evidence ranged from "very low" to "moderate" across study designs and health outcomes using GRADE. In conclusion, the available evidence supports that earlier sleep timing and regularity in sleep patterns with consistent bedtimes and wake-up times are favourably associated with health. (PROSPERO registration no.: CRD42019119534.) Novelty This is the first systematic review to examine the influence of sleep timing and sleep consistency on health outcomes. Later sleep timing and greater variability in sleep are both associated with adverse health outcomes in adults. Regularity in sleep patterns with consistent bedtimes and wake-up times should be encouraged.

Keywords: bedtime; catch-up sleep; directives; décalage horaire social; guidelines; heure du coucher; heure du réveil; midpoint of sleep; point médian du sommeil; public health; régularité du sommeil; santé publique; sleep regularity; sleep variability; social jetlag; sommeil de rattrapage; variabilité du sommeil; wake-up time.

Publication types

• Systematic Review
• Aging / physiology
• Aging / psychology
• Cardiometabolic Risk Factors
• Health Status*
• Mental Health
• Quality of Life
• Sedentary Behavior
• Sleep Hygiene* / physiology
• Time Factors

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here .

Loading metrics

Open Access

Peer-reviewed

Research Article

Examining relationships between sleep posture, waking spinal symptoms and quality of sleep: A cross sectional study

Roles Conceptualization, Formal analysis, Investigation, Methodology, Writing – original draft, Writing – review & editing

* E-mail: [email protected]

Affiliations School of Allied Health, Faculty of Health Science, Curtin University, Bentley, Western Australia, Australia, Esperance Physiotherapy, Esperance, Western Australia, Australia

Roles Formal analysis

Affiliation School of Allied Health, Faculty of Health Science, Curtin University, Bentley, Western Australia, Australia

Roles Formal analysis, Supervision, Writing – review & editing

• Doug Cary, 
• Angela Jacques, 
• Kathy Briffa

• Published: November 30, 2021
• https://doi.org/10.1371/journal.pone.0260582
• Peer Review
• Reader Comments

Introduction

Research with a focus on sleep posture has been conducted in association with sleep pathologies such as insomnia and positional obstructive sleep apnoea. Research examining the potential role sleep posture may have on waking spinal symptoms and quality of sleep is however limited. The aims of this research were to compare sleep posture and sleep quality in participants with and without waking spinal symptoms.

Fifty-three participants (36 female) were, based on symptoms, allocated to one of three groups; Control ( n = 20, 16 female), Cervical ( n = 13, 10 female) and Lumbar ( n = 20, 10 female). Participants completed an online survey to collect general information and patient reported outcomes and were videoed over two consecutive nights to determine sleep posture using a validated classification system including intermediate sleep postures.

Participants in the symptomatic groups also reported a lower sleep quality than the Control group. Compared to Control group participants, those in the Cervical group had more frequent posture changes (mean (SD); 18.3(6.5) versus 23.6(6.6)), spent more time in undesirable/provocative sleep postures (median IQR; 83.8(16.4,105.2) versus 185.1(118.0,251.8)) minutes and had more long periods of immobility in a provocative posture, (median IQR: 0.5(0.0,1.5) versus 2.0 (1.5,4.0)). There were no significant differences between the Control and Lumbar groups in the number of posture changes (18.3(6.5) versus 22.9(9.1)) or the time spent in provocative sleep postures (0.5(0.0,1.5) versus 1.5(1.5,3.4)) minutes.

This is the first study using a validated objective measure of sleep posture to compare symptomatic and Control group participants sleeping in their home environment. In general, participants with waking spinal symptoms spent more time in provocative sleep postures, and experienced poorer sleep quality.

Citation: Cary D, Jacques A, Briffa K (2021) Examining relationships between sleep posture, waking spinal symptoms and quality of sleep: A cross sectional study. PLoS ONE 16(11): e0260582. https://doi.org/10.1371/journal.pone.0260582

Editor: Matias Noll, Instituto Federal Goiano, BRAZIL

Received: November 7, 2020; Accepted: November 14, 2021; Published: November 30, 2021

Copyright: © 2021 Cary et al. This is an open access article distributed under the terms of the Creative Commons Attribution License , which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: This study was conducted with the approval of the Human Research Ethics Committee at Curtin University. This approval did not extent to public sharing of the dataset. Requests for permission to access the data should be sent to the Human Research Ethics Committee, Curtin University ( [email protected] ).

Funding: I would like to acknowledge support from the Australian Government (Research Training Program Scholarship) and Curtin University (Manuscript Writing Grant) (DC). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Sleep is considered essential for human mental and physical recovery [ 1 – 3 ], with some people going to bed in pain, to wake recovered. Nonetheless, a proportion of the population who are asymptomatic when retiring wake with spinal symptoms and others with existing spinal symptoms wake with exacerbations of their retiring spinal symptoms [ 4 – 7 ].

Spinal symptoms are common and mostly occur in the cervical and lumbar regions, with a one-year point prevalence of 30 to 50% for cervical pain [ 8 ] and 38% for lumbar pain [ 9 ]. The prevalence of both cervical and lumbar pain has increased markedly (cervical 21.1% and lumbar 17.3%) over the past 25 years, and these rates are expected to continue rising [ 10 ]. Other types of symptoms like stiffness and bothersomeness, still important to patients, are less well documented.

It has been postulated that poor sleep posture during the night may be responsible for the production of waking cervical [ 11 – 13 ] and lumbar symptoms [ 14 ]. It was determined in young military recruits, that 33% had their most intense spinal pain during sleep hours or on waking and that for 50% of the recruits, the spinal pain was significant enough to cause disruption to their sleep routine [ 4 ].

Anecdotal and theoretical evidence suggests that mechanical loads induced by some sleep postures, like prone, may provoke spinal pain [ 6 , 15 ]. Collagen containing tissues like ligaments, intervertebral discs and capsules, undergo predictable mechanical and viscoelastic changes such as creep, hysteresis and fatigue failure in response to a single sustained load or to repeated loads [ 16 , 17 ]. Loads sustained for greater than 10 minutes and repeated loads causing 3% or greater elongation, have resulted in collagenous tissue micro-damage. Micro-damage has been associated with an increased expression of pro-inflammatory cytokines in animal studies [ 18 ]. Muscle spasms associated with sustained flexion or rotation spinal postures have been reported in both human [ 19 , 20 ] and animal studies [ 18 ]. It therefore seems plausible, that sleep postures sustained for greater than 10 minutes or repeated, may cause micro damage and result in spinal symptoms.

Sleep posture has also been associated with sleep quality. Poor quality of sleep is subjectively determined by delayed sleep onset, more awakenings after sleep onset, increased total wake time, and poor continuity of sleep [ 21 , 22 ]. Therefore, factors like sleep posture that provokes spinal pain, potentially causing increased total wake time, could impact on sleep quality. Poor sleep quality is significantly associated with adverse health outcomes for adults [ 23 , 24 ] and is predictive of musculoskeletal pain in pre-adolescents, adolescents, young adults [ 25 – 28 ] and adults without pain [ 24 ]. For this reason, it is important to identify any factor that potentially could adversely affect an individual’s ability to maintain an asleep state.

The aims of this research were to determine whether there were differences in sleep posture and sleep quality in participants with and without waking spinal symptoms.

This study was approved by the Curtin University Human Research Ethics Committee (HR 140/2014). Approval to share data beyond the investigators was not obtained. Written informed consent was obtained from all participants. The study was registered with the Australian New Zealand Clinical Trials Registry on 4/07/2014 (ACTRN12614000708651). The subject in Fig 1 has given written informed consent.

• PPT PowerPoint slide
• PNG larger image
• TIFF original image

Sustained postures like rotation, have been identified as causing tissue microdamage and muscle spasms.

https://doi.org/10.1371/journal.pone.0260582.g001

Recruitment occurred in Esperance, a rural town of Western Australia through word of mouth, recruitment posters, radio interviews, letters to possible referrers and newspaper advertisements. Volunteers were asked screening questions in a phone or face to face interview to determine eligibility for the study. Volunteers who were younger than 18 years or older than 46 years were excluded. The younger age was for legal consent reasons and the upper age to minimise the chances of confounding factors like increasing severity of spinal degenerative changes [ 29 ]. Volunteers with medical conditions such as severe osteoarthritis, spinal stenosis, oesophageal reflux and late stage pregnancy or using devices such as breathing apparatus that prevented them from sleeping in all postures were excluded [ 30 ]. Those with co-existing medically diagnosed inflammatory conditions or unremitting pain (e.g., rheumatoid arthritis, neuropathic pain) were also excluded. Volunteers using medically prescribed hypnotic or relaxant medications would have been excluded as these medications can alter frequency of posture changes during sleep, but none were excluded for this reason. A total of 53 participants (36 female) with predominately morning symptoms of pain, stiffness or bothersomeness were recruited over a period of 2.5 years. Using a process of best fit, eligible volunteers were allocated to one of the three groups, Control ( n = 20, 16 female), Cervical ( n = 13, 10 female) and Lumbar ( n = 20, 10 female). Participants with spinal pain, stiffness or bothersomeness greater than or equal to 3 out of 10 on a numerical rating scale (NRS) [ 31 ], that occurred four or more times per month and decreased within 60 minutes of waking, were allocated into a symptomatic group (Cervical or Lumbar) dependent on their self-reported dominant area of symptoms. Participants without symptoms, or with symptoms less frequently than four times per month, or less than 3 out of 10, were allocated to the Control group. Each symptomatic group was individually compared with the Control group.

Outcome measures

Due to the vanguard nature of this study, a broad range of pain, disability, sleep and quality of life patient reported outcome measures were collected to facilitate better understanding of possible relationships between sleep posture and waking spinal symptoms. Participants in all three groups were emailed a link to a Survey Monkey questionnaire, enabling the online completion of baseline information (e.g., age, gender, body mass index, medications, level of education, and self-reported sleep posture) and patient reported outcome measures. Data for the following patient reported outcome measures was collected.

Numerical rating scale.

Numerical rating scales for waking pain, stiffness, bothersomeness and quality of sleep in the prior 2 weeks. Higher scores indicated increased symptoms for pain, stiffness and bothersomeness, whereas a higher score indicated a better quality of sleep.

Neck Disability Index (NDI).

A 10-item, self-reported questionnaire measuring cervical disability. In a group of patients with non-specific neck pain, a minimal clinically important difference of 3.5 points best distinguished those patients who were clinically improved from those who were not [ 32 ].

Spine Functional Index (SFI).

A 10-item, 2 weeks recall whole of spine functional measure, the SFI-10 demonstrated high criterion validity with the Functional Rating Index (r = .87), equivalent internal consistency (α = .91) and a single-factor structure in patients with spinal pain referred to physiotherapy clinics by medical practitioners [ 33 , 34 ].

Roland Morris Disability Questionnaire (RMQ).

A 24-item, immediate recall self-reported lumbar disability measure, found to be reliable, valid and responsive to change over time. Higher scores indicated greater disability.

Hospital Anxiety and Depression Scale (HADS).

A 14-item, two domain assessment widely used to identify cases of anxiety and depression in non-psychiatric hospital clinics for adults greater than 16 years of age [ 35 ]. Each item is scored on a 4-point Likert scale (0–3) generating anxiety and depression scores ranging from 0 to 21.

Short Form—36 (SF-36).

The Short Form-36 is a well-validated measure of health status. Version 2 (1 week recall) was used with Australian normative data [ 36 , 37 ] to calculate two summary scores (i.e., Physical Component and Mental Component Score). A higher score indicates a better health status.

Insomnia Sleep Index (ISI).

A 7-item, 2 weeks recall self-reported scale designed to assess the nature, severity and impact of insomnia and to monitor treatment response in adults [ 38 , 39 ]. Scale development included a heterogenous group of patients with insomnia secondary to pain conditions [ 40 ].

Pittsburgh Sleep Quality Index (PSQI).

A 19-item, 7-domain, 2 weeks recall self-reported questionnaire which examines subjective sleep quality, with global scores ranging from 0 to 21. A global PSQI score greater than 5 is considered a poor sleeper and yielded a diagnostic sensitivity of 89.6% and specificity of 86.5% (kappa = .75, p < .001) in distinguishing between good and poor sleepers. Higher scores indicate poorer sleep quality [ 41 , 42 ].

Sleep posture.

The classification of sleep posture into supine, supportive side lying, provocative side lying and prone ( Fig 1 ), and the method used to collect sleep posture data has been previously described [ 43 ]. The validity and reliability of this method has also been previously reported [ 44 ]. In brief, two IR cameras were installed in the participant’s bedroom on portable stands. One camera was placed at the foot end of the bed, and the other centrally over the bed. These IR cameras were cable-connected to a digital data recorder. A monitor was connected to the data recorder to optimise camera viewing fields. Data collection automatically commenced at 2000hrs and stopped at 0800hrs. Participants were encouraged to maintain their normal sleeping routines.

After two consecutive nights video equipment was retrieved and recordings were reviewed. Head, trunk and leg positions were noted and the overall sleep posture was recorded as supine, supportive side lying, provocative side lying or prone. Each posture change was manually recorded relative to the time stamp and rounded up to the next half minute. For example, 1 to 29 seconds became a half minute and 31 to 59 seconds became a full minute. To be recorded, a posture needed to be sustained for at least 1 minute. Head movements from neutral to right or left rotation, without a change in trunk position were recorded, but were not considered a new posture because of no major change in load on the spine. If participants moved from right to left supported side lying or provocative side lying, this was recorded as a new posture, due to the major change in body posture. Sustained posture intervals of 30 minutes or greater were described as long periods of postural immobility Researchers have indicated that long periods of postural immobility are an indication of sleep stability [ 45 , 46 ]. Exploring the idea that some sleep postures may be provocative of spinal symptoms, our study sought to not only measure the frequency of long periods of postural immobility (these will be referred to as standard long periods of postural immobility), but also the posture in which the long periods of postural immobility occurred and the number of 30-minute periods for each long periods of postural immobility (referred to as actual long periods of postural immobility). That is, a posture held for 65 minutes would be recorded as one standard long period of postural immobility but two (30-minute) actual long periods of postural immobility.

In addition to the collection of video data, participants completed a Morning After Questionnaire each morning after being videoed, to score pain, stiffness, bothersomeness and quality of sleep on a NRS in relation to the prior night.

Sample size

A priori sample size calculations were based on data collected in a pilot study [ 43 ]. It was calculated a sample of 30 in each group (i.e., Control, Cervical, Lumbar) would have a power of 99%, assuming a two tailed p-value of .05 to detect a large effect (0.8). Further, a sample of 60 people with symptoms would have sufficient size to detect a minimal clinically important change in pain of 1.5 points on a NRS assuming a standard deviation (SD) of 2 points, with a power of 99% and a minimal clinically important change of 5 points on the SF36 summary scales, assuming a SD of 10 points with a power of 96% following an intervention. The sample sizes of 30 for each group were not achieved, rather final sample sizes for each group were Control (20), Cervical (13) and Lumbar (20).

Data analysis

Questionnaires were scored according to published scoring algorithms or instructions provided by developers. Sleep data from Night 1 and Night 2 were averaged prior to further analyses. Statistical analysis was performed using IBM ® SPSS ® v24.0. All data were checked for outliers by visual inspection of boxplots or population pyramids. Outlying data points were checked for data entry errors and measurement errors.

On initial review of raw video data and before data analysis commenced, outliers within groups were identified. It became apparent, that while some participants were by definition sleeping in provocative side lying (i.e., top thigh advanced forward of the bottom thigh, Fig 1 ), because of the use of pillows and/or duvet to support the upper thigh, participants technically did not induce spinal rotation or extension. For this reason, the provocative side lying sleep postures of three participants were reclassified to supportive side lying. Genuinely unusual values were rare and retained in the analyses.

Descriptive statistics were presented as count and percentage, mean ( M ) and standard deviation ( SD ) or median and interquartile range (IQR). A range of sleep posture variables were extracted from the video data to examine possible relationships with waking spinal symptoms; specifically, the percentage of time and the total amount of time spent in each sleep posture (i.e., supine, supportive side lying, provocative side lying and prone). For some analyses, postures were grouped into supportive (i.e., supine and supportive side lying) and provocative (i.e., provocative side lying and prone), based upon plausible spinal load. The distribution of the data was examined using numerical (Shapiro-Wilk test) and graphical (visual examination of Histograms and Normal Q-Q Plots) methods. Achieving a normal distribution was not possible for most Control group variables, particularly patient reported outcome variables, as most participants in this group reported low or no symptom levels. After outliers and normality assumptions were checked, homogeneity of variance was checked using the Levene statistic for normally distributed data ( p > .05 significant). Between group comparisons (Cervical versus Control and Lumbar versus Control) were undertaken using a one-way analysis of variance (ANOVA) statistic (F), or for non-normally distributed data, a Mann-Whitney U test (U). A chi-square test was used to compare categorical variables between groups. A p < .05 (two tailed where appropriate) was considered significant for all analyses.

Group characteristics

The age of participants ranged from 18 to 45 years, with the largest group of participants in the 41 to 45 years range. Overall there were more female than male participants, with 16 females in the Control group and 10 in both the Cervical and Lumbar groups. There were no significant differences between groups in distribution of age, gender, education or BMI scores ( Table 1 ). Participants nominated the types of medications and supplements they were currently using. Approximately the same percentage in each group used none, one to two and three or more medications or supplements. The types of medications and supplements used in each group are detailed in Table 1 .

https://doi.org/10.1371/journal.pone.0260582.t001

Patient reported outcomes

Participants in the Cervical group had significantly worse pain, stiffness, and bothersomeness on waking than the Control group ( Table 2 ). They also recorded significantly poorer scores in all of the patient reported outcome questionnaires except the SF-36 MS ( Table 2 ).

https://doi.org/10.1371/journal.pone.0260582.t002

Participants in the Lumbar group had significantly worse pain, stiffness, and bothersomeness on waking than the Control group ( Table 2 ). They also recorded significantly poorer scores in all of the patient reported outcome questionnaires except the HADS—Depression ( Table 2 ).

Comparison of sleep posture variables

Compared with the Control group, participants in the Cervical group spent a greater percentage of the night in provocative side lying and combined provocative sleep postures. When time in each posture was expressed in absolute values (minutes), the Cervical group spent, on average, twice as long in provocative side lying ( Table 3 ). Participants in the Cervical group changed their sleep postures more frequently than the Control group and spent more of their long periods of postural immobility in provocative sleep postures ( Table 3 ).

https://doi.org/10.1371/journal.pone.0260582.t003

There was no statistically significant difference (p = .052) between the time participants in the Lumbar group spent in provocative sleep postures compared with the Control group ( Table 3 ). Nor were there any statistically significant differences between the Lumbar and Control groups in postural immobility (p > .07) ( Table 3 ).

Comparison of sleep quality variables

With respect to sleep quality, both symptom groups self-reported significantly lower sleep quality than the Control group using the numerical rating scale ( Table 3 ). Scores from the Insomnia Sleep Index and Pittsburgh Sleep Quality Index also indicated poorer sleep quality in both the Cervical and Lumbar groups ( Table 3 ). For the Cervical group however, the average Insomnia Sleep Index score remained within the ‘no clinically significant insomnia’ band of scores whereas the Lumbar group, on average, would be classified as having ‘subthreshold insomnia’. Both groups would be classified as ‘poor sleepers’ using the Pittsburgh Sleep Quality Index classification.

Volunteers for this study were allocated into one of two symptomatic groups (Cervical or Lumbar) or a Control group based on the location, duration and intensity of their self-reported symptoms. As would be expected with classification of this nature, the symptomatic groups had greater morning symptom scores and decrements in the various patient reported outcomes measured (summarised in Tables 1 and 2 ).

The findings of the study are consistent with the theories posed prior to the study; that people with spinal pain would spend more of the night in provocative sleep postures and would have lower sleep quality than a Control group. Interpretation of these findings must be in the context of the study design and limited sample size. Given the cross-sectional study design, it is not possible to be sure whether any of the variables of interest are causal or whether their presence is simply coincidental. Small studies are more vulnerable to Type II errors than larger studies.

The participants with morning symptoms of neck pain or stiffness slept differently to those without morning symptoms. In comparison with the Control group, participants in the Cervical group spent significantly more time in provocative sleep postures. These results are similar to an epidemiological study examining waking cervical symptoms and sleep posture [ 6 ]. In that study, participants who reported prone as their dominant sleep posture also reported the highest percentage of waking cervical symptoms. An interesting consideration is whether it is the amount of time or the percentage of time spent in provocative sleep postures that is more likely to provoke symptoms. With regards to the Cervical group, we found both the percentage time and the total time spent in provocative postures was greater than the Control group. A study conducted on feline spines points towards the amount of time as being important and once this threshold is passed, recovery of tissue takes proportionally much longer [ 47 ].

We examined the frequency of posture shifts from the stand point of plausible tissue load. One possible reason to change posture more frequently would be to offload pain sensitive structures more likely to be aggravated by sleep postures such as prone and provocative side lying. Spinal and capsular ligaments are highly innervated [ 48 ] and have been shown to produce pro-inflammatory cytokines following sustained or repeated loading in feline studies [ 18 ]. Pain free adults of mixed age and gender have been noted to change posture approximately 12 to 20 times per night [ 49 – 51 ]. This frequency of posture shifts is reported to double in those describing themselves as poor sleepers [ 22 ]. In comparison to our Control group, the Cervical group did experience a significantly higher frequency of posture shifts.

We did not find any differences in Actual or Standard long periods of postural immobility between the Control and Symptomatic groups using the common long period of postural immobility definition of a 30-minute interval. Given other researchers have noted that spinal tissue creep occurs within 10 minutes [ 19 , 20 ], to be more sensitive with respect to plausible spinal tissue loading, perhaps the concept of a long period of postural immobility needs to be modified to a shorter time-period. To our knowledge, we are the first group to examine the relevance of long periods of postural immobility in relation to provocative and supportive sleep postures. In comparison to participants in the Control group, participants in the Cervical group spent more long periods of postural immobility in provocative sleep postures. This result runs contrary to the theory that long periods of postural immobility are postures of comfort and therefore sustained for longer periods of time. It is however consistent with our theory, that sustained sleeping in provocative postures could cause waking spinal symptoms. Rather than reflecting postures of comfort, it may be that postures adopted for long periods are postures of habit. These postures may be amenable to educational interventions like posture retraining [ 52 , 53 ].

Participants in the Lumbar group spent three times as long as those in the Control group in provocative postures, but results were not statistically significant. Visual examination of the data in Table 3 reveals that values for the Lumbar group consistently fell between those for the Control and Cervical groups, never reaching statistical significance. It is possible that we did not have sufficient statistical power to identify a difference between the Control and Lumbar groups causing a Type II error. It is also possible that, as a result of our recruitment criteria, the differences between our Control and Lumbar groups were reduced. For example, at enrolment, 45% of participants in the Control group nominated having some low back pain in the prior 2 weeks, however their level of reported pain was insufficient to meet the criteria for allocation into the Lumbar group.

In this study, sleep quality was measured in three ways; numerical rating scale for sleep quality over the prior 2 weeks, the Pittsburgh Sleep Quality Index and the Insomnia Sleep Index. The latter measures are commonly used in the sleep literature and quantify different aspects of sleep quality. Good quality sleep has been associated with a side lying sleep posture [ 6 ]. However, given the wide variety of side lying sleep postures, it is likely that some side lying postures may place adverse loads on spinal tissues and not be conducive to good quality sleep.

The relationship between sleep quality and pain has historically been considered bidirectional. More recent research points to sleep quality as being the antecedent factor [ 54 ] and in the insomnia research literature, poor sleep quality is considered predictive of new pain onsets and exacerbation of existing pain [ 55 , 56 ]. If sleep posture influences sleep quality, then optimising sleep posture could potentially reduce spinal pain via two separate mechanisms. Firstly, by reducing collagenous spinal tissue load and injury associated with creep, and secondly, by improving sleep quality.

Neck and low back pain are global health problems and the challenge to identify risk factors has been highlighted [ 57 , 58 ]. Determining modifiable risk factors could assist in the identification of individuals predisposed to spinal symptoms and assist in the development of appropriate education and prevention strategies. In a recent systematic review examining risk factors for first episode neck pain, the most significant physical risk factor was an awkward, sustained posture [ 57 ]. When examining trigger events (i.e., brief exposures) that precipitate acute low back pain, symptom onset was most common in the morning [ 59 ]. The timing implicates sleep posture as a possible factor and passive collagenous restraints, (e.g., IVD, ligament, joint capsule) as being the tissues most likely affected [ 60 ]. Sleep posture was not explored in either study. Further, the lack of research focus on spinal symptoms associated with sleep posture was highlighted in a recent scoping review, in which only four studies were found to address these topics [ 61 ].

Currently, it is not known if sleep posture is a risk factor for acute onset or recurrent spinal symptoms, but this study has demonstrated that participants with symptoms of cervical pain and stiffness in the morning spent more of the night in provocative sleep postures. It is plausible that a similar association exists for people with symptoms in the low back, however this was not confirmed in this study. Both symptomatic groups had poorer sleep quality. Future exploration of the effects of provocative sleep posture and potential benefits of sleep posture education and modification seem justified.

• 1. Cabot S, Beckham N. Tired of not Sleeping? Camden: WHAS Pty Ltd; 2005.
• 2. Helmanis L. Sleep Better Naturally. London: Carlton Books; 2006.
• View Article
• PubMed/NCBI
• Google Scholar
• 4. Desouzart G, Vilar E, Melo F, Matos R, editors. Human bed interaction: A methodology and tool to measure postural behavior during sleep of the air force military. 3rd International Conference on Design, User Experience, and Usability; 2014 June; Heraklion, Greece.
• 7. Kraemer J. Intervertebral disk diseases: Causes, diagnosis, treatment and prophylaxis. 3 ed. New York: Thieme Medical Publishers; 2011.
• 16. Adams M, Bogduk N, Burton K, Dolan P. The biomechanics of back pain. 2 ed. Edinburgh: Churchill Livingstone; 2006.
• 17. PitchingNow. Stress strain curve. PitchingNow; 2014.
• 36. Marin T, Taylor A, Gill T. A population profile of auality of life in South Australia: Population norms for 2008 using the SF36 version 2. In: Unit PRaOS, editor. Adelaide: South Australian Department of Health; 2009.
• 50. Chen Y, Guo Y, Shen L, Liu S, editors. The quantitative effects of mattress and sleep postures on sleep quality. International Asia Conference on Industrial Engineering and Management Innovation; 2013; Berlin Heidelberg: Springer-Verlag.

• Search Menu
• Supplements
• Cohort Profiles
• Education Corner
• Author Guidelines
• Submission Site
• Open Access
• About the International Journal of Epidemiology
• About the International Epidemiological Association
• Editorial Team
• Editorial Board
• Advertising and Corporate Services
• Journals Career Network
• Self-Archiving Policy
• Dispatch Dates
• Contact the IEA
• Journals on Oxford Academic
• Books on Oxford Academic

Article Contents

Conclusions, conflict of interest, appendix: hierarchical bayesian model for case-control studies (front vs non-front).

• < Previous

Infant sleeping position and the sudden infant death syndrome: systematic review of observational studies and historical review of recommendations from 1940 to 2002

• Article contents
• Figures & tables
• Supplementary Data

Ruth Gilbert, Georgia Salanti, Melissa Harden, Sarah See, Infant sleeping position and the sudden infant death syndrome: systematic review of observational studies and historical review of recommendations from 1940 to 2002, International Journal of Epidemiology , Volume 34, Issue 4, August 2005, Pages 874–887, https://doi.org/10.1093/ije/dyi088

• Permissions Icon Permissions

Background Before the early 1990s, parents were advised to place infants to sleep on their front contrary to evidence from clinical research.

Methods We systematically reviewed associations between infant sleeping positions and sudden infant death syndrome (SIDS), explored sources of heterogeneity, and compared findings with published recommendations.

Results By 1970, there was a statistically significantly increased risk of SIDS for front sleeping compared with back (pooled odds ratio (OR) 2.93; 95% confidence interval (CI) 1.15, 7.47), and by 1986, for front compared with other positions (five studies, pooled OR 3.00; 1.69–5.31). The OR for front vs the back position was reduced as the prevalence of the front position in controls increased. The pooled OR for studies conducted before advice changed to avoid front sleeping was 2.95 (95% CI 1.69–5.15), and after was 6.91 (4.63–10.32). Sleeping on the front was recommended in books between 1943 and 1988 based on extrapolation from untested theory

Conclusions Advice to put infants to sleep on the front for nearly a half century was contrary to evidence available from 1970 that this was likely to be harmful. Systematic review of preventable risk factors for SIDS from 1970 would have led to earlier recognition of the risks of sleeping on the front and might have prevented over 10 000 infant deaths in the UK and at least 50 000 in Europe, the USA, and Australasia. Attenuation of the observed harm with increased adoption of the front position probably reflects a ‘healthy adopter’ phenomenon in that families at low risk of SIDS were more likely to adhere to prevailing health advice. This phenomenon is likely to be a general problem in the use of observational studies for assessing the safety of health promotion.

Sudden unexpected unexplained infant death, now known as sudden infant death syndrome (SIDS), was recognized as a major cause of infant death in the UK and USA throughout the 20th century. At the start of the 20th century, such deaths were attributed to overlying, particularly by drunken mothers. 1 By the 1940s, as more deaths were investigated by autopsy, pathologists realized that few deaths were due to maternal overlying, and alternative mechanisms for ‘accidental mechanical suffocation’ were sought. In 1944, Abramson, a pathologist in New York State, noted that two-thirds of infants dying from mechanical suffocation were found face down, contrary to the usual sleeping position for infants at the time. 2 His observations, which were corroborated by reports in the UK and Australia 3 , 4 led to a health promotion campaign that recommended avoidance of the front position. 5

The campaign was short-lived. In 1945, a paediatrician, Woolley, rejected Abramson's hypothesis of suffocation on the front based on experiments in which he had covered babies' faces with layers of blankets. 6 He reported that the oxygen content of the air breathed by the babies only fell when they were covered with a rubber sheet and that babies moved if breathing was obstructed. He also criticized the explanation of suffocation because it ‘instilled guilt and self-incrimination in parents’.

Emergence of alternative explanations for death, such as unrecognized infection 4 , 7 , 8 inhalation of vomit 9 and hypersensitivity reaction to inhaled milk, 10 further strengthened the argument against the suffocation hypothesis and highlighted the need for data on risk factors. The first published case–control study was started in 1956 in the USA, 11 and in 1958, a similar study in the UK was the first to measure infant sleeping position in SIDS victims and live control babies. 12 At around the same time, it became increasingly common to advocate sleeping on the front. We now know that front sleeping is a major cause of SIDS. We wanted to know whether systematic review of the evidence could have reversed this harmful advice sooner or whether variation in the association between sleeping on the front and SIDS was consistent with recommendations at the time. We did a systematic review and meta-analysis of the effect of front and side sleeping on the risk of SIDS, and an historical review of recommendations on infant sleeping position in books and pamphlets on infant care available in the UK between 1940 and 2002. We focussed on how the strength of the evidence for a harmful effect of front sleeping changed before and after advice changed in favour of avoidance of the front position. We hypothesized that the effect of the front position on SIDS might differ depending on whether health advice favoured front or not as families that adopt health advice are likely to be at lower risk of SIDS.

Historical review

We reviewed recommendations on infant sleeping position in books or pamphlets available in the UK from 1940 to 2002. We chose 1940 to include a period before the front position was widely advocated. We searched the Modern Medicine Collection at the Wellcome Trust library, and, because of a lack of more recent texts, the British Medical Association library from 1965 to 2002. We included any book or pamphlet that referred to the care of normal term infants aged <6 months, and mentioned infant sleeping position. Searches used the library indexing system for books on infant care and we also searched electronically using terms for paediatric, parent, and baby (details of search strategy available from authors).

One reviewer (S.S. or M.H.), assessed whether texts met the inclusion criteria and prepared a hard copy file with the extract and book title but not the date of publication. A second reviewer (R.G.) categorized the recommendation as favouring front, back, side, or non-front position(s), or neutral if all or none were implicitly or explicitly favoured. A second reviewer (S.S.), independently categorized one-third of the texts and there was complete agreement with the first reviewer.

Systematic review

We included any case–control or cohort study that compared the risk of SIDS in infants sleeping on their front, side, or back. Studies had to be based on SIDS infants and live healthy control infants from the same community. We searched for any comparative study of infant sleeping position and SIDS in MEDLINE (1966–2002) and EMBASE (1980–2002), using a detailed search strategy (available from the authors), and reference lists of review articles, a PhD thesis on the history of SIDS, 13 and included studies. Abstracts were scanned by one reviewer (S.S., M.H., or R.G.), and full texts of potentially eligible studies retrieved. R.G. and S.P. jointly extracted data from included studies.

Data quality

We used data on the position in which the infant was placed to sleep before death or interview, or if lacking, data on usual position, or position found. If usual position was measured at multiple ages, we used results closest to 3 months of age. We recorded the method of selection of cases and controls, matching criteria, if any, and whether data collection methods differed in cases and controls.

Our primary aim was to compare the risk of SIDS in infants sleeping front and back. As some studies did not separately report side and back positions, we also compared front with non-front positions. However, grouping side with back will attenuate the observed risk associated with the front position if the side position is also harmful. We therefore calculated odds ratios (ORs) for SIDS associated with sleeping front vs back, front vs non-front, and side vs back.

To avoid confounding, we used the unadjusted matched OR if reported. Otherwise we calculated the unmatched OR. 14 Because studies differed in their design, populations, and methods, we used a random effects model in which it is assumed that the observed ORs are sampled from a common distribution around a mean effect with variance measured by the heterogeneity parameter. We estimated 95% confidence intervals (CIs) and considered a P -value <0.05 as statistically significant. Heterogeneity in the OR for SIDS was assessed by the chi-squared test (Q-test) and quantified using I 2 which reflects the proportion of variation that is not due to sampling error. 15 The possibility of publication bias was evaluated using funnel plots and the Egger and Begg tests. 16 , 17

We determined the year at which there was a statistically significant association between front or side sleeping positions and SIDS by using a cumulative meta-analysis based on year of publication as described by Lau. 18 The overall heterogeneity was used in the calculation of the CIs for the cumulative OR at every step using a random effects model. We applied recursive cumulative meta-analysis to examine the direction and magnitude of the relative changes in the cumulative evidence as a function of the cumulative sample size. 19 , 20 At the end of every information period j , the ratio (cumulative OR j )/(cumulative OR j + 1 ) was assessed and compared with unity. If larger than one, this was interpreted as a ‘move’ of the evidence towards defining the front position as more harmful than in the previous information period.

To explore potential sources of heterogeneity we initially used conventional meta-regression to determine an association with variables previously suggested. 21 In a univariate model, we first determined the effect of the position recorded in cases (before death, usual, or after death), year of publication, recruitment year (measured as the mid-point between start and end of recruitment), matching criteria for controls and cases, and country and continent of study. The combined effect on heterogeneity of the variables found to be significant in the univariate analysis was estimated in a multivariate meta-regression model. We extended the meta-regression analyses to examine the hypothesis that the prevalence of front sleeping in control infants is associated with heterogeneity. This is because parents who put their babies to sleep in the front position when advised not to, might have a different risk of SIDS than parents who do so when front sleeping is the norm (similarly for the side position).

The OR for front vs any other position can be written as using logOR = logit P (front|case) − logit P (front|control) and the prevalence of front sleeping is estimated in the controls as P (front) = P (front|control). Consequently, regression of logOR to P (front) will be biased by regression to the mean.

To overcome this we fitted a hierarchical model similar to that described by Thompson et al . to model background risk in randomized controlled trials. 22 , 23 As the studies were case-control rather than trials, we made some modifications to the methods (see Appendix).

We retained in the model any factors that were statistically significantly associated with heterogeneity in the conventional meta-regression, and assessed the extent to which the factors included in the model explained the variation between studies by measuring the change in the heterogeneity parameter. If factors included in the model explained heterogeneity, the heterogeneity parameter (variance in the random effects) would be expected to get smaller. This model was fitted using Markov chain Monte Carlo methods within a Bayesian framework. The analysis was conducted using Intercooled Stata 8.2 (Stata Corp., College Station, TX), R 1.9.1 (R Foundation for statistical computing, Vienna) and Winbugs 1.4.1.

Table 1 summarizes the recommendations made in 83 texts that met the inclusion criteria (details available from the authors). From 1940 to the mid-1950s, texts favoured the back or side positions and only one, in 1943, recommended the front position. From 1954 until 1988, a substantial proportion of texts consistently favoured front sleeping, although many also favoured the side and back. The sudden shift in favour of front sleeping is best illustrated by ‘Baby and Child Care’ by Dr Benjamin Spock who recommended the back position in his 1955 edition, and the front position in 1956. 24 In his 1958 edition, he argued ‘If he vomits, he's more likely to choke on the vomitus. Also, he tends to keep his head turned to the same side—usually toward the centre of the room. This may flatten the side of his head.’ Many authors repeated these arguments. Others argued that front sleeping reduced wind, 25 , 26 coughing due to mucus, 27 and made respiration easier. 26 Suffocation was considered to be possible only if the baby was very weak. 26 These views were not universal. In editions of his textbook in 1945, 1950, and 1959, Nelson stated that ‘position during sleep is relatively unimportant, but should be changed often to prevent moulding of the cranium’. 28 – 30 Others were less equivocal. One author recommended in 1953, ‘Sleeping on his abdomen never should be permitted because of the danger of suffocating’. 31 In 1966, another warned ‘Very small babies should never be left alone lying on their tummies. This is an American fashion to strengthen the back, but we think the dangers of suffocation are not sufficiently remote to justify it.’ 32

Recommended infant sleeping position in books on infant care

Books written by Dr Benjamin Spock.

No texts favoured the front position after 1988. From the mid-1950s to 1990, many texts continued to recommend the side position, but few advocated sleeping on the back. In the early 1990s, most texts recommended the side position or simply advised against front sleeping, but apart from one text in 1990, the back position was not consistently advocated until 1995.

Of the 2897 abstracts scanned, and the 206 full text articles retrieved, 40 studies met the inclusion criteria ( Figure 1 and Table 2 ). Four further studies were excluded ( Figure 1 ). No randomized controlled trials were found. All 40 included studies provided data on front vs non-front positions, but only 24 studies separately recorded back and side positions. Of the 40 studies, 23 (and 15/24 reporting side and back positions) included some degree of matching of controls with cases. Of these, unadjusted matched ORs were available for 9/23 studies (and for 7/15 reporting side and back positions). 33 , 33 – 44 For one study, we derived pooled ORs from data reported for separate ethnic groups. 37 All studies were case–control except for one cohort study reported in two stages. This resulted in data for 2 years of the study (15 SIDS victims) being included twice in the cumulative meta-analyses. 45 , 46 Repeated use of the same data was avoided for all the other studies except for Mitchell 2 1999 (details in Table 2 ). No substantial evidence was found for publication bias for any of the sleeping position comparisons either by examining the funnel plots or applying the Egger or Begg tests (lowest P -value = 0.103).

Flow diagram to show results of searches for the systematic review of comparative studies of infant sleeping position and SIDS

Characteristics of included studies

NK; not known.

The results for side, back and front positions were published in 1972 as a histogram. 89 Actual figures have been supplied by the author.

Studies after this point included populations advised to avoid front sleeping. Four studies were excluded 67 , 109 – 111

Data for separate centres provided by the author and then pooled.

1 = position placed to sleep before death or interview; 2 = usual position; 3 = position found.

There was a statistically significantly higher risk of death associated with the front position whether compared with the back ( Figure 2a ) or non-front positions ( Figure 2c ). There was a weak association between the side position and the risk of SIDS, which was marginally worse than back ( Figure 2e ).

Forest plots show ORs for SIDS and pooled OR for comparisons of (a) front vs back; (c) front vs non-front; and (e) side vs back sleeping positions. Figures 2b, d, and f , depict the cumulative meta-analyses for front vs back (b), front vs non-front (d), and side vs back (f). Studies ordered by publication date

The cumulative meta-analyses showed that the association between death and the front position compared with back had become statistically significant by 1970, after the first two case–control studies (cumulated OR 2.93; 95%CI 1.15–7.47; Figure 2b ). When front was compared with non-front, the association was not statistically significant until 1986, after inclusion of five studies (cumulated OR 3.00; 1.69–5.31; Figure 2d ). Recursive meta-analysis showed that the relative magnitude of the cumulative OR for front vs back changed by up to 22% from one publication year to the next between 1986 and 1996, but remained stable (maximum change 4%) when studies published after 1996 were included (results not shown). After 1996, populations included in these studies were advised to use the side or back positions (see Table 2 ).

Substantial heterogeneity was detected in all three datasets as shown in the forest plots ( Figures 2a, c, and e ) and reflected in the highly significant Q statistic and high values for I 2 (83% for front vs back, 89% for front vs non-front, and 73% for side vs back). In the conventional meta-regression the only significant factor was the year of recruitment, with later studies associated with an increased OR for SIDS in all three comparisons. The results of extending the meta-regression to include the prevalence of front or side positions in control babies are shown in Table 3 . For front compared with back, the prevalence of the front position was the only factor that was significantly associated with heterogeneity. As the prevalence of the front position in control babies increased, the OR for SIDS decreased. For front vs non-front positions, there was little evidence that prevalence of front position or year of recruitment explained heterogeneity. Finally, in the comparison of side vs back, only the prevalence of the side position was associated with a reduction in the OR, but had little effect on heterogeneity.

Results of meta-regression adjusted for prevalence of front or side position in control infants a

Restricted to 38 case–control studies.

Credibility interval: there is a 95% probability that the true value lies within the 95% credibility interval.

Absolute reduction in between-study variance between the crude meta-analysis model and the meta-regression model.

The front sleeping position was recommended from 1943 to 1988 although the first text to advise against front sleeping was not published until 1992. The safest position—on the back—was recommended sporadically during the 1980s but not consistently until 1995. However, by 1970 the pooled evidence from two studies showed that the risk of SIDS was statistically significantly higher for babies on the front than on the back. The harmful effect of front sleeping was lowest when the prevalence of the front position in control babies was highest.

A detailed historical analysis of why clinicians recommended that infants sleep on the front is beyond the scope of this study. From the reasons given for advocating front sleeping, 47 there is no clear evidence that the back position increases the risk of crying, 46 , 48 – 50 inhalation of vomit, or colic. 46 , 50 , 51 However, in the short term, sleeping on the front is associated with increased motor development, 52 , 53 rounder head shape, 54 nappy rash, 49 , 50 and pyloric stenosis. 55 Front sleeping is also associated with longer sleep duration, 46 , 48 , 50 probably by reducing physiological control of respiratory, cardiovascular and autonomic control mechanisms, and arousal during sleep. 56

Our analyses identified five factors that may have contributed to the delayed recognition of the risks of front sleeping: the paucity of published studies between 1970 and 1986; the marked heterogeneity among studies; the relationship between the prevalence of front sleeping and year of recruitment and the size of the OR; and grouping of the comparator as non-front in some studies. Finally, many authors interpreted the front position as just one of a number of factors associated with SIDS and did not systematically review results from previous studies. 12 , 57 – 59 The first overview of studies on the effect of sleeping position was published by Beal in 1988. 60

It was striking that no studies were published on the effect of sleeping position between 1970 and 1986. Although several investigators collected data on sleeping position during the 1970s and early 1980s, their findings were not published until 1986 or later. 33 , 58 , 59 , 61 – 65 Sleeping position may have been disregarded because the front position was not directly compared with the back, and the results of Frogatt and Carpenter were not combined. In addition, Frogatt 66 questioned the validity of his results because they were only statistically significant when the usual sleeping position was compared, not if the position in which the SIDS victim was found was used. Bergman, 67 may have further deterred research on sleeping position after finding that 85% of SIDS victims in a large US study were found on the front, and claiming, without reporting any control data, that this was similar to the community.

The lack of research attention on infant sleeping position between 1970 and 1986 contrasts with the increasing incidence of SIDS, and the steep increase in the proportion of infants sleeping front in several industrialized countries ( Figures 3a, b, and c , and Figure 4 ). 68 – 75 In the UK, the increase in SIDS incidence was attributed to diagnostic transfer—deaths previously classified as due to respiratory causes being classified as SIDS, which became a registrable cause only in 1971. However, there was concern that, while all other causes of infant deaths had declined during the 1970s and 1980s, SIDS and respiratory deaths combined had remained static. 75 , 76 Clear evidence that SIDS incidence had truly increased and was not due to diagnostic transfer was not published until the 1990s ( Figure 3c ). 68 , 69 , 74 In contrast, the decline in incidence following advice to avoid front sleeping in the ‘Back to Sleep’ campaigns ( Figures 3a, b, and c ) was rapid and undeniable, providing the strongest evidence to date for a harmful effect of the front position. SIDS incidence fell by 50–70% in numerous countries, in association with a fall in front sleeping. ( Figures 3a, b, and c ) 75 , 77 , 78

(a) Post-neonatal SIDS mortality (infant deaths due to SIDS after the first month of life) in England and Wales 1974–1998 (Arrows depict publication of Avon SIDS study July 1990, 34 and UK National ‘Back to Sleep’ campaign, November 1991); (b) SIDS incidence (deaths in the first year per 1000 live births) in Australia, New Zealand and the USA; and (c) SIDS incidence in Sweden, Norway, and The Netherlands

Prevalence of the front position among healthy infants based on controls in included studies and community studies from 1958 to 1998 33 , 34 , 39 , 44 , 57 , 63 , 67 – 69 , 84 , 89 – 92

A crude estimate of the number of babies who died in England and Wales owing to harmful health advice can be made by assuming that the rate of post neonatal SIDS would have remained at 0.6/1000 live births, the rate in the year after the government's ‘Back to Sleep’ campaign. From 1974, when SIDS was routinely used as a cause of death, until 1991, there were 11 000 excess deaths, or nearly 12 extra babies dying each week. However, the number of excess deaths is highest in the USA, where the prevalence of front sleeping was higher for longer than in any other country 48 , 79 ( Figure 4 ). In the USA, rest of Europe, and Australasia, at least 50 000 excess deaths were attributable to harmful health advice.

We found substantial heterogeneity in the association between sleeping position and SIDS that was partly explained by the prevalence of the front (or side) position in control infants, and to a lesser extent, year of recruitment. In an era when front sleeping was the norm, parents who placed infants on the back were likely to have had socioeconomic characteristics that put them at high risk for SIDS, thereby diminishing the observed protective effect of the back position. 80 , 81 Conversely, when prevailing advice was to avoid front sleeping, characteristics in those that did not take up this advice exaggerated the observed harmful effect of the front position. In other words, increased uptake of advice by families otherwise at low risk of SIDS produced a ‘healthy adopter’ effect that diminished evidence of harm. An alternative explanation is biased reporting of the position considered to be harmful by parents of SIDS victims. Another possibility is that studies showing an adverse effect of the sleeping position advocated at the time were less likely to be written about and published.

The effect of the era of health advice is best illustrated by comparing the pooled ORs for front vs back positions, before and after advice changed. For studies published between 1965 12 and 1995, 37 the pooled OR was 2.95 (95% CI: 1.69–5.15, studies); thereafter the pooled OR was 6.91 (4.63–10.32). In the example of SIDS, a statistically significant association was still detectable because the OR was relatively large. However, these findings raise a general message for the evaluation of potentially harmful health advice that uptake by people at low risk of adverse outcomes could completely obscure evidence of harm. 82

The fact that much heterogeneity between studies remained unexplained may be partly owing to difficulties in accurately measuring study characteristics. For example, we could not adequately measure the potential for reporting bias, which may have contributed to the relatively low OR for SIDS in three studies because staff responsible for recommending the front position also selected control babies and/or collected the data. 61 , 64 , 65 , 83 A second factor in three studies, all conducted in the USA, may be the close matching of controls with cases based on age, hospital of birth, and ethnic group. 37 , 40 , 84 If there had been uniform adoption of health advice within these communities, such close matching may have biased the association towards the null effect. In the first two of these studies, close matching, combined with the high prevalence of front sleeping, may have contributed to the relatively weak associations observed. 37 , 84 Factors contributing to heterogeneity may also differ according to the era of health advice. This may partly explain differences between our results and a previous meta-analysis, restricted to studies published before 1990, that found that country of study, date of publication, matching, and position reported were associated with heterogeneity when sleeping front was compared with non-front. 21

It is unusual for health advice to have such a profound effect on mortality and to detect such tragic effects from health advice that is not based on evidence of effectiveness. Had systematic reviews been common practice in the early 1970s, parents, professionals, and policy makers would have been aware of the cumulative effect of the front position on SIDS at least 15 years earlier than they were in 1988. Even if the results had been judged insufficient to change practice, they should have stimulated earlier publication of further studies.

Others have similarly highlighted the delayed introduction of effective treatment that could have been avoided if systematic review and meta-analysis had been used to summarize the accumulated evidence from randomized controlled trials. 85 , 86 Interpretation of systematic reviews of observational studies is more difficult owing to the potential for bias and spurious precision. 87 , 88 In particular, our results show that observational studies of health advice can be confounded by a ‘healthy adopter’ phenomenon that can diminish or obscure adverse effects of harmful health advice. All these problems are compounded when examining multiple risk factors. Nevertheless, when randomized controlled trials are lacking or not feasible, systematic review of observational studies is essential to guide policy and practice.

R.G. coordinated one of the included studies. 34 None of the other authors have any conflict of interest.

Advice to put infants to sleep on the front for nearly a half century was contrary to evidence available from 1970 that this was likely to be harmful.

Systematic review of preventable risk factors for SIDS from 1970 would have led to earlier recognition of the risks of sleeping on the front and might have prevented over 10 000 infant deaths in the UK and at least 50 000 in Europe, the USA, and Australasia.

Attenuation of the observed harm with increased adoption of the front position probably reflects a ‘healthy adopter’ phenomenon in that families at low risk of SIDS were more likely to adhere to prevailing health advice.

We thank Iain Chalmers, Julian Higgins, and Jan van der Meulen for comments on an earlier draft of this article. Chris Hiley gave material from her PhD on the History of SIDS and Sima Patel acted as second reviewer for data extraction and helped with preliminary analyses. Bob Carpenter provided unpublished data for two of his studies. The Foundation for the Study of Infant Deaths allowed access to their archives. We thank the reviewers of an earlier version for their constructive comments, and Adèle Engelberts for providing incidence data for The Netherlands. All data are available on the web: http://www.ich.ucl.ac.uk/ich/html/academicunits/paed_epid/cebch/

Westcott WW. Inebriety in women and the overlaying of infants. The British Journal of Inebriety 1903 ; 4i : 65 –68.

Abramson H. Accidental mechanical suffocation in infants. The Journal of Pediatrics 1944 ; 25 : 404 –13.

Bowden KM. Sudden death or alleged accidental suffocation in babies. The Medical Journal of Australia 1950 ; 1 : 65 –72.

Davison H. Accidental infant suffocation. BMJ 1945 ; 2 : 251 –52.

New York State Department of Health. Merchanical suffocation: leading cause of accidental death in early infancy. Health News 1945 ; 22 : 33 –34.

Woolley PV. Mechanical suffocation during infancy. A comment on its relation to the total problem of sudden death. The Journal of Pediatrics 1945 ; 26 : 572 –75.

Werne J, Garrow I. Sudden deaths of infants allegedly due to merchanical suffocation. Am J Public Health 1947 ; 37 : 675 –87.

Werne J, Garrow I. Sudden apparently unexplained death during infancy. I Pathologic findings in infants found dead. Am J Pathol 1953 ; 29 : 633 –53.

Polson CJ, Price DE. Suffocation by milk feeds. Lancet 1948 ; 251 : 343 .

Camps FE, Parish WE, Barrett AM, Coombs RRA, Gunther M. Hypersensitivity to milk and sudden death in infancy. Lancet 1960 ; 276 : 1106 –10.

Strimer R, Adelson L, Oseasohn R. Epidemiologic features of 1,134 sudden, unexpected infant deaths. A study in the Greater Cleveland Area from 1956 to 1965. JAMA 1969 ; 209 : 1493 –97.

Carpenter RG, Shaddick CW. Role of infection, suffocations and bottle-feeding in cot death: an analysis of some factors in the histories of 110 cases and their controls. Br J Prev Soc Med 1965 ; 19 : 1 –7.

Hiley CMH. Back to Sleep: Cot death and Infant Care 1987–1994 . Cambridge: University of Cambridge, 1995 .

Greenland S, Rothman KJ. Measures of effect and measures of association. In: Rothman KJ, Greenland S (eds). Modern Epidemiology . Philadelphia: Lippincott-Raven, 1998 , pp. 47–64.

Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003 ; 327 : 557 –60.

Egger M, Davey SG, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997 ; 315 : 629 –34.

Begg CB, Mazumdar M. Operating characteristics of a rank correlation test for publication bias. Biometrics 1994 ; 50 : 1088 –101.

Lau J, Antman EM, Jimenez-Silva J, Kupelnick B, Mosteller F, Chalmers TC. Cumulative meta-analysis of therapeutic trials for myocardial infarction. N Engl J Med 1992 ; 327 : 248 –54.

Trikalinos TA, Churchill R, Ferri M et al . Effect sizes in cumulative meta-analyses of mental health randomized trials evolved over time. J Clin Epidemiol 2004 ; 57 : 1124 –30.

Ioannidis JP, Contopoulos-Ioannidis DG, Lau J. Recursive cumulative meta-analysis: a diagnostic for the evolution of total randomized evidence from group and individual patient data. J Clin Epidemiol 1999 ; 52 : 281 –91.

Dwyer T, Couper D, Walter SD. Sources of heterogeneity in the meta-analysis of observational studies: the example of SIDS and sleeping position. J Clin Epidemiol 2001 ; 54 : 440 –47.

Sharp SJ, Thompson SG. Analysing the relationship between treatment effect and underlying risk in meta-analysis: comparison and development of approaches. Stat Med 2000 ; 19 : 3251 –74.

Thompson SG, Smith TC, Sharp SJ. Investigating underlying risk as a source of heterogeneity in meta-analysis. Stat Med 1997 ; 16 : 2741 –58.

Spock B. Baby and Child Care . London: The Bodley Head, 1958 .

Illingworth RS. The Normal Child . London: J & A Churchill Ltd, 1968 .

Shepard KS. Care of the Well Baby . Philadelphia: JP LIppincott Company, 1960 .

Bax M, Bernal J. Your Child's First Five Years . London: Heinemann Health Books, 1974 .

Nelson WE. Mitchell-Nelson Textbook of Pediatrics . London: WB Saunders, 1945 .

Nelson WE. Mitchell-Nelson Textbook of Pediatrics . London: WB Saunders Company, 1950 .

Nelson WE. Mitchell-Nelson Textbook of Pediatrics . London: WB Saunders Company, 1959 .

Zabriskie L. Mother and Baby in Pictures . London: JB Lippincott Company, 1953 .

Gordon M, Gordon R. A Baby in the House. A Guide to Practical Parenthood . London: Heinemann, 1966 .

Nicholl JP, O'Cathain A, McGlashan ND. Sleeping positions and SIDS. Lancet 1988 ; 3322 : 106 .

Fleming PJ, Gilbert R, Azaz Y et al . Interaction between bedding and sleeping position in the sudden infant death syndrome: a population based case–control study. BMJ 1990 ; 301 : 85 –89.

Ponsonby AL, Dwyer T, Gibbons LE, Cochrane JA, Wang YG. Factors potentiating the risk of sudden infant death syndrome associated with the prone position. N Engl J Med 1993 ; 329 : 377 –82.

Markestad T, Skadberg B, Hordvik E, Morild I, Irgens LM. Sleeping position and sudden infant death syndrome (SIDS): effect of an intervention programme to avoid prone sleeping. Acta Paediatr 1995 ; 84 : 375 –78.

Klonoff-Cohen HS, Edelstein SL. A case–control study of routine and death scene sleep position and sudden infant death syndrome in Southern California. JAMA 1995 ; 273 : 790 –94.

Taylor JA, Krieger JW, Reay DT, Davis RL, Harruff R, Cheney LK. Prone sleep position and the sudden infant death syndrome in King County, Washington: a case–control study. J Pediatr 1996 ; 128 : 626 –30.

Fleming PJ, Blair PS, Bacon C et al . Environment of infants during sleep and risk of the sudden infant death syndrome: results of 1993–5 case–control study for confidential inquiry into stillbirths and deaths in infancy. Confidential Enquiry into Stillbirths and Deaths Regional Coordinators and Researchers. BMJ 1996 ; 313 : 191 –95.

Hauck FR, Moore CM, Herman SM et al . The contribution of prone sleeping position to the racial disparity in sudden infant death syndrome: the Chicago Infant Mortality Study. Pediatrics 2002 ; 110 : 772 –80.

Carpenter RG, Irgens LM, Blair PS et al . Sudden unexplained infant death in 20 regions in Europe: case control–study. Lancet 2004 ; 363 : 185 –91.

Brooke H, Gibson A, Tappin D, Brown H. Case–control study of sudden infant death syndrome in Scotland, 1992–5. BMJ 1997 ; 314 : 1516 –20.

McGarvey C, McDonnell M, Chong A, O'Regan M, Matthews T. Factors relating to the infant's last sleep environment in sudden infant death syndrome in the Republic of Ireland. Arch Dis Child 2003 ; 88 : 1058 –64.

Oyen N, Markestad T, Skaerven R et al . Combined effects of sleeping position and prenatal risk factors in sudden infant death syndrome: the Nordic Epidemiological SIDS Study. Pediatrics 1997 ; 100 : 613 –21.

Dwyer T, Ponsonby AL, Newman NM, Gibbons LE. Prospective cohort study of prone sleeping position and sudden infant death syndrome. Lancet 1991 ; 337 : 1244 –47.

Dwyer T, Ponsonby AL, Couper D, Cochrane J. Short-term morbidity and infant mortality among infants who slept supine at 1 month of age—a follow-up report. Paediatr Perinat Epidemiol 1999 ; 13 : 302 –15.

Anon. Prone or supine? BMJ 1961 ; 6 : 1304 .

Brackbill Y, Douthitt TC, West H. Psychophysiologic effects in the neonate of prone versus supine placement. J Pediatr 1973 ; 82 : 82 –84.

Keitel HG, Cohn R, Harnish D. Diaper rash, self-inflicted excoriations, and crying in full-term newborn infants kept in the prone or supine position. J Pediatr 1960 ; 57 : 884 –86.

Hunt L, Fleming P, Golding J. Does the supine sleeping position have any adverse effects on the child? I. Health in the first six months. The ALSPAC Study Team. Pediatrics 1997 ; 100: E11 .

Byard RW, Beal SM. Gastric aspiration and sleeping position in infancy and early childhood. J Paediatr Child Health 2000 ; 36 : 403 –5.

Dewey C, Fleming P, Golding J. Does the supine sleeping position have any adverse effects on the child? II. Development in the first 18 months.ALSPAC Study Team. Pediatrics 1998 ; 101 : E5 .

Davis BE, Moon RY, Sachs HC, Ottolini MC. Effects of sleep position on infant motor development. Pediatrics 1998 ; 102 : 1135 –40.

Persing J, James H, Swanson J, Kattwinkel J. Prevention and management of positional skull deformities in infants. American Academy of Pediatrics Committee on Practice and Ambulatory Medicine, Section on Plastic Surgery and Section on Neurological Surgery. Pediatrics 2003 ; 112 : 199 –202.

Persson S, Ekbom A, Granath F, Nordenskjold A. Parallel incidences of sudden infant death syndrome and infantile hypertrophic pyloric stenosis: a common cause? Pediatrics 2001 ; 108 : E70 .

Galland BC, Taylor BJ, Bolton DP. Prone versus supine sleep position: a review of the physiological studies in SIDS research. J Paediatr Child Health 2002 ; 38 : 332 –8.

Froggatt P. Epidemiological aspects of the Northern Ireland study. In: Bergman AB, Beckwith JB, Ray CG (eds). Sudden Infant Death Syndrome. Proceedings of the Second International Conference on Causes of Sudden Death in Infants . Seattle: University of Washington Press, 1970 , pp. 32–46.

McGlashan ND. Sudden infant deaths in Tasmania, 1980–1986: a seven year prospective study. Soc Sci Med 1989 ; 29 : 1015 –26.

Cameron MH, Williams AL. Development and testing of scoring systems for predicting infants with high-risk of sudden infant death syndrome in Melbourne. Aust Paediatr J 1986 ; 22 (Suppl. 1): 37 –45.

Beal S. Sleeping position and SIDS. Lancet 1988 ; 2 : 512 .

Tonkin SL. Epidemiology of cot deaths in Auckland. NZ Med J 1986 ; 99 : 324 –26.

Beal SM. Sudden infant death syndrome: epidemiological comparisons between South Australia and communities with a different incidence. Aust Paediatr J 1986 ; 22 (Suppl. 1): 13 –16.

de Jonge GA, Engelberts AC, Koomen-Liefting AJM, Kostense PJ. Cot death and prone sleeping position in The Netherlands. BMJ 1989 ; 298 : 722 .

Tonkin S,.Hassell I. Infant sleeping position and cot death. Aust Paediatr J 1989 ; 25 : 376 –77.

Hassall IB, VandenBerg M. Infant sleep position: a New Zealand survey. NZ Med J 1985 ; 98 : 97 –99.

Froggatt P, Lynas MA, MacKenzie G. Epidemiology of sudden unexpected death in infants (‘cot death’) in Northern Ireland. Br J Prev Soc Med 1971 ; 25 : 119 –34.

Bergman AB, Ray CG, Pomeroy MA, Wahl PW, Beckwith JB. Studies of the sudden infant death syndrome in King County, Washington. 3. Epidemiology. Pediatrics 1972 ; 49 : 860 –70.

Engelberts AC, de Jonge GA, Kostense PJ. An analysis of trends in the incidence of sudden infant death in The Netherlands 1969–89. J Paediatr Child Health 1991 ; 27 : 329 –33.

Irgens LM, Markestad T, Baste V, Schreuder P, Skjaerven R, Oyen N. Sleeping position and sudden infant death syndrome in Norway 1967–91. Arch Dis Child 1995 ; 72 : 478 –82.

Hogberg U, Bergstrom E. Suffocated prone: the iatrogenic tragedy of SIDS. Am J Public Health 2000 ; 90 : 527 –31.

Mitchell EA, Ford RP, Taylor BJ et al . Further evidence supporting a causal relationship between prone sleeping position and SIDS. J Paediatr Child Health 1992 ; 28 (Suppl. 1): S9 –S12.

Taylor BJ. A review of epidemiological studies of sudden infant death syndrome in southern New Zealand. J Paediatr Child Health 1991 ; 27 : 344 –48.

Beal SM. Sudden infant death syndrome in South Australia 1968–97. Part I: changes over time. J Paediatr Child Health 2000 ; 36 : 540 –47.

Mitchell EA. International trends in postneonatal mortality. Arch Dis Child 1990 ; 65 : 607 –09.

Gilbert R. The changing epidemiology of SIDS. Arch Dis Child 1994 ; 70 : 445 –49.

Golding J, Limerick S, MacFarlane A. Variation of incidence with time, place, and age of infant. In: Golding J, Limerick S, MacFarlane A (eds). Sudden Infant Death: Patterns, Puzzles, and Problems . Shepton Mallet, UK: Open Books Publishing Ltd, 1985 , pp. 22–42.

Willinger M, Hoffman HJ, Hartford RB. Infant sleep position and risk for sudden infant death syndrome: report of meeting held January 13 and 14, 1994, National Institutes of Health, Bethesda, MD. Pediatrics 1994 ; 93 : 814 –19.

McKee M, Fulop N, Bouvier P et al . Preventing sudden infant deaths—the slow diffusion of an idea. Health Policy 1996 ; 37 : 117 –35.

Pollack HA, Frohna JG. Infant sleep placement after the back to sleep campaign. Pediatrics 2002 ; 109 : 608 –14.

Spencer N, Logan S. Sudden unexpected death in infancy and socioeconomic status: a systematic review. J Epidemiol Community Health 2004 ; 58 : 366 –73.

Ponsonby AL, Dwyer T, Kasl SV, Couper D, Cochrane JA. Correlates of prone infant sleeping position by period of birth. Arch Dis Child 1995 ; 72 : 204 –08.

Ebrahim S, Smith GD. Health promotion for coronary heart disease: past, present and future. Eur Heart J 1998 ; 19 : 1751 –57.

Engelberts AC. Cot Death in The Netherlands: An Epidemiological Study . Amsterdam: VU University Press, 1991 .

Hoffman HJ, Hillman LS. Epidemiology of the sudden infant death syndrome: maternal, neonatal, and postneonatal risk factors. Clin Perinatol 1992 ; 19 : 717 –37.

Crowley P. Antenatal corticosteroids—current thinking. BJOG 2003 ; 110 (Suppl. 20): 77 –78.

Antman EM, Lau J, Kupelnick B, Mosteller F, Chalmers TC. A comparison of results of meta-analyses of randomized control trials and recommendations of clinical experts. Treatments for myocardial infarction. JAMA 1992 ; 268 : 240 –48.

Egger M, Schneider M, Davey SG. Spurious precision? Meta-analysis of observational studies. BMJ 1998 ; 316 : 140 –44.

Ioannidis JP, Haidich AB, Pappa M et al . Comparison of evidence of treatment effects in randomized and nonrandomized studies. JAMA 2001 ; 286 : 821 –30.

Carpenter RG. Sudden and unexpected deaths in infancy (cot death). In: Camps FE, Carpenter RG (eds). Sudden and Unexpected Deaths in Infancy (Cot Death) . Bristol. John Wright, 1972 , pp. 7–15.

Wigfield RE, Fleming PJ, Berry PJ, Rudd PT, Golding J. Can the fall in Avon's sudden infant death rate be explained by changes in sleeping position? BMJ 1992 ; 304 : 282 –83.

Hiley CM, Morley CJ. Risk factors for sudden infant death syndrome: further change in 1992–3. BMJ 1996 ; 312 : 1397 –98.

Willinger M, Hoffman HJ, Wu KT et al . Factors associated with the transition to nonprone sleep positions of infants in the United States: the National Infant Sleep Position Study. JAMA 1998 ; 280 : 329 –35.

Senecal J, Roussey M, Defawe G, Delahaye M, Piquemal B. Procubitus et mort subite inattendue du nourrison. Arch Fr Pediatr 1987 ; 44 : 131 –36

Beal SM, Finch CF. An overview of retrospective case-control studies investigating the relationship between prone sleeping position and SIDS. J Paediatr Child Health 1991 ; 27 : 334 –39.

Lee M, Davies DP, Chan YF. Prone or supine for preterm babies? Lancet 1988 ; 1 : 1332 .

Bouvier-Colle MH, Varnoux V, Hausherr E. Revue bibliographique des etudes sur la mort subite en relation avec la position de sommeil chez le nourrisson. Proceedings of Reunion du Groupe d'Etudes de Langue Francaise sur la mort subite du nourrisson. Nice. France. 1990 .

Mitchell EA, Scragg R, Stewart AW et al . Results from the first year of the New Zealand cot death study. NZ Med J 1991 ; 104 : 71 –76.

Jorch G, Schmidt TS, Bajanowski T et al . Risk factors for sudden infant death (SID): epidemiologic study of two German districts 1990–1992 epidemiologische risikofaktoren des plotzlichen kindstods. Ergebnisse der westfalischen kindstodstudie 1990–1992. Monatsschr Kinderheilkd 1994 ; 142 : 45 –51.

Gormally S, Matthews TG. Sleep position and SIDS in Irish infants. Ir Med J 1994 ; 87 : 58 .

Andersen M, Arnestad M, Rognum TO, Vege A. Krybbedod i ostlandsregionen 1984–92. En kartlegging av risikofaktorer. [Crib death in the eastern regions of Norway 1984–1992. A survey of risk factors]. Tidsskr Nor Laegeforen 1995 ; 115 : 34 –37.

Mitchell EA, Tuohy PG, Brunt JM et al . Risk factors for sudden infant death syndrome following the prevention campaign in New Zealand: a prospective study. Pediatrics 1997 ; 100 : 835 –40.

Schellscheidt J, Ott A, Jorch G. Epidemiological features of sudden infant death after a German intervention campaign in 1992. Eur J Pediatr 1997 ; 156 : 655 –60.

Kleemann WJ, Schlaud M, Fieguth A, Hiller AS, Rothamel T, Troger HD. Body and head position, covering of the head by bedding and risk of sudden infant death (SID). Int J Legal Med 1999 ; 112 : 22 –26.

Skadberg BT, Morild I, Markestad T. Abandoning prone sleeping: effect on the risk of sudden infant death syndrome. J Pediatr 1998 ; 132 : 340 –43.

L'Hoir MP, Engelberts AC, van Well GT et al . Case-control study of current validity of previously described risk factors for SIDS in The Netherlands. Arch Dis Child 1998 ; 79 : 386 –93.

Mitchell EA, Thach BT, Thompson JM, Williams S. Changing infants' sleep position increases risk of sudden infant death syndrome. New Zealand Cot Death Study. Arch Pediatr Adolesc Med 1999 ; 153 : 1136 –41.

Dwyer T, Ponsonby AL, Gibbons LE, Newman NM. Prone sleeping position and SIDS: evidence from recent case-control and cohort studies in Tasmania. J Paediatr Child Health 1991 ; 27 : 340 –43.

Toro K, Sotonyi P. Distribution of prenatal and postnatal risk factors for sudden infant death in Budapest. Scand J Prim Health Care 2001 ; 19 : 178 –80.

Saternus KS. Plötzlicher Kindstod-eine Folge der Bauchlage? Festschrift Professor Leithoff . Heidelberg: Kriminalistik Verlag, 1985 , pp. 67–81.

Kahn A, Blum D, Hennart P et al . A critical comparison of the history of sudden-death infants and infants hospitalised for near-miss for SIDS. Eur J Pediatr 1984 ; 143 : 103 –07.

Nelson EA, Taylor BJ, Mackay SC. Child care practices and the sudden infant death syndrome. Aust Paediatr J 1989 ; 25 : 202 –04.

Email alerts

Citing articles via, looking for your next opportunity.

• About International Journal of Epidemiology
• Recommend to your Library

Affiliations

• Online ISSN 1464-3685
• Copyright © 2024 International Epidemiological Association
• About Oxford Academic
• Publish journals with us
• University press partners
• What we publish
• New features  
• Open access
• Institutional account management
• Rights and permissions
• Get help with access
• Accessibility
• Advertising
• Media enquiries
• Oxford University Press
• Oxford Languages
• University of Oxford

Oxford University Press is a department of the University of Oxford. It furthers the University's objective of excellence in research, scholarship, and education by publishing worldwide

• Copyright © 2024 Oxford University Press
• Cookie settings
• Cookie policy
• Privacy policy
• Legal notice

This Feature Is Available To Subscribers Only

Sign In or Create an Account

This PDF is available to Subscribers Only

For full access to this pdf, sign in to an existing account, or purchase an annual subscription.

Advertisement

Standardized framework to report on the role of sleeping position in sleep apnea patients

• Sleep Breathing Physiology and Disorders • Review
• Published: 11 January 2021
• Volume 25 , pages 1717–1728, ( 2021 )

Cite this article

• M. J. L. Ravesloot   ORCID: orcid.org/0000-0001-5645-5811 1 ,
• P. E. Vonk 2 ,
• J.T. Maurer 3 ,
• A. Oksenberg 4 &
• N. de Vries 1 , 5 , 6  

1087 Accesses

16 Citations

Explore all metrics

Sleep apnea is a multifactorial illness which can be differentiated in various physiological phenotypes as a result of both anatomical and non-anatomical contributors (e.g., low respiratory arousal threshold, high loop gain). In addition, the frequency and duration of apneas, in the majority of patients with OSA, are influenced by sleeping position. Differences in characteristics between non-positional patients (NPP) and positional patients (PP) suggest another crucial phenotype distinction, a clinical phenotype focusing on the role of sleeping position on sleep apnea. Since this clinical phenotype distinction has therapeutic implications, further research is necessary to better understand the pathophysiology behind this phenotypic trait and to improve management of PP. Therefore, we suggest a standardized framework that emphasizes the role of sleeping position when reporting clinical and research data on sleep apnea.

We identified 5 key topics whereby a standardized framework to report on the role of sleeping position would be of added value: (1) sleep study data, (2) anatomical, morphological and physiological factors, (3) drug-induced sleep endoscopy (DISE) findings, (4) sleep apnea management, and (5) effectiveness versus efficacy of positional therapy in sleep apnea management. We performed a literature search to identify evidence to describe and support the rationale behind these 5 main recommendations.

In this paper, we present the rationale behind this construct and present specific recommendations such as reporting sleep study indices (disease severity) and sleep time spent in various sleeping positions. The same is suggested for DISE findings and effect of treatment. Sleep study indices (disease severity), anatomical, morphological, and physiological factors in sleep apnea patients should be reported separately for PP and NPP.

Applying these suggestions in future research will improve patient care, assist in better understanding of this dominant phenotype, and will enhance accurate comparisons across studies and future investigations.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price includes VAT (Russian Federation)

Instant access to the full article PDF.

Rent this article via DeepDyve

Institutional subscriptions

Similar content being viewed by others

Sleep position and obstructive sleep apnea (OSA): Do we know how we sleep? A new explorative sleeping questionnaire

Positional obstructive sleep apnea.

Positional therapy in the management of positional obstructive sleep apnea—a review of the current literature

Data availability.

Not applicable.

Kezirian EJ, Weaver EM, Criswell MA, de Vries N, Woodson BT, Piccirillo JF (2011) Reporting results of obstructive sleep apnea syndrome surgery trials. Otolaryngol Head Neck Surg 144(4):496–499

PubMed   PubMed Central   Google Scholar  

Edwards BA, Redline S, Sands SA, Owens RL (2019) More than the sum of the respiratory events: personalized medicine approaches for obstructive sleep apnea. Am J Respir Crit Care Med 200(6):691–703. https://doi.org/10.1164/rccm.201901-0014TR

Article   PubMed   PubMed Central   Google Scholar  

Taranto-Montemurro L, Messineo L, Wellman A (2019) Targeting endotypic traits with medications for the pharmacological treatment of obstructive sleep apnea. A review of the current literature. Journal of. Clin Med 8(11):1846

CAS   Google Scholar  

Eckert DJ, White DP, Jordan AS, Malhotra A, Wellman A (2013) Defining phenotypic causes of obstructive sleep apnea. Identification of novel therapeutic targets. Am J Respir Crit Care Med 188(8):996–1004

Eckert DJ (2018) Phenotypic approaches to positional therapy for obstructive sleep apnoea. Sleep Med Rev 37:175–176. https://doi.org/10.1016/j.smrv.2017.06.007

Article   PubMed   Google Scholar  

Oksenberg A, Gadoth N (2018) Positional patients (PP) and non positional patients (NPP) are two dominant phenotypes that should be included in the phenotypic approaches to obstructive sleep apnea. Sleep Med Rev 37:173–174

PubMed   Google Scholar  

Ravesloot MJ, Frank MH, van Maanen JP, Verhagen EA, de Lange J, de Vries N (2015) Positional OSA part 2: retrospective cohort analysis with a new classification system (APOC). Sleep Breath 20:881–888. https://doi.org/10.1007/s11325-015-1206-y

Oksenberg A, Silverberg DS, Arons E, Radwan H (1997) Positional vs nonpositional obstructive sleep apnea patients: anthropomorphic, nocturnal polysomnographic, and multiple sleep latency test data. Chest 112(3):629–639

CAS   PubMed   Google Scholar  

Richard W, Kox D, den Herder C, Laman M, van Tinteren H, de Vries N (2006) The role of sleep position in obstructive sleep apnea syndrome. Euro Archiv Oto-Rhino-Laryngol Euro Federat Oto-Rhino-Laryngologic Societ (EUFOS) : Affil German Soc Oto-Rhino-Laryngol Head Neck Surg 263(10):946–950. https://doi.org/10.1007/s00405-006-0090-2

Article   Google Scholar  

Mador MJ, Kufel TJ, Magalang UJ, Rajesh SK, Watwe V, Grant BJ (2005) Prevalence of positional sleep apnea in patients undergoing polysomnography. Chest 128(4):2130–2137. https://doi.org/10.1378/chest.128.4.2130

Oksenberg A, Gadoth N, Töyräs J, Leppänen T (2019) Prevalence and characteristics of positional obstructive sleep apnea (POSA) in patients with severe OSA. Sleep Breath 24(2):551–559

Oksenberg A, Khamaysi I, Silverberg DS, Tarasiuk A (2000) Association of body position with severity of apneic events in patients with severe nonpositional obstructive sleep apnea. Chest 118(4):1018–1024

Cartwright RD (1984) Effect of sleep position on sleep apnea severity. Sleep 7(2):110–114

Bignold JJ, Mercer JD, Antic NA, McEvoy RD, Catcheside PG (2011) Accurate position monitoring and improved supine-dependent obstructive sleep apnea with a new position recording and supine avoidance device. J Clin Sleep Med 7(4):376–383. https://doi.org/10.5664/jcsm.1194

Marklund M, Persson M, Franklin KA (1998) Treatment success with a mandibular advancement device is related to supine-dependent sleep apnea. Chest 114(6):1630–1635

Permut I, Diaz-Abad M, Chatila W, Crocetti J, Gaughan JP, D’Alonzo GE, Krachman SL (2010) Comparison of positional therapy to CPAP in patients with positional obstructive sleep apnea. J Clin Sleep Med 6(3):238–243

Joosten SA, O’donoghue FJ, Rochford PD, Barnes M, Hamza K, Churchward TJ, Berger PJ, Hamilton GS (2014) Night-to-night repeatability of supine-related obstructive sleep apnea. Ann Am Thoracic Soc 11(5):761–769

Google Scholar  

Heinzer R, Petitpierre NJ, Marti-Soler H, Haba-Rubio J (2018) Prevalence and characteristics of positional sleep apnea in the HypnoLaus population-based cohort. Sleep Med 48:157–162

Kim KT, Cho YW, Kim DE, Hwang SH, Song ML, Motamedi GK (2016) Two subtypes of positional obstructive sleep apnea: supine-predominant and supine-isolated. Clin Neurophysiol 127(1):565–570

Duce B, Kulkas A, Langton C, Töyräs J, Hukins C (2017) Amsterdam positional OSA classification: the AASM 2012 recommended hypopnoea criteria increases the number of positional therapy candidates. Sleep Breath 21(2):411–417. https://doi.org/10.1007/s11325-016-1432-y

van Kesteren ER, van Maanen JP, Hilgevoord AA, Laman DM, de Vries N (2011) Quantitative effects of trunk and head position on the apnea hypopnea index in obstructive sleep apnea. Sleep 34(8):1075–1081. https://doi.org/10.5665/sleep.1164

Walsh JH, Maddison KJ, Platt PR, Hillman DR, Eastwood PR (2008) Influence of head extension, flexion, and rotation on collapsibility of the passive upper airway. Sleep 31(10):1440–1447

Janssen H, Verbraecken JA, Pevernagie D, Overeem S (2015) Positional central sleep apnea. In: Positional therapy in obstructive sleep apnea. Springer, Berlin, pp 209–219

Benoist L, Vonk P, de Vries N, Janssen H, Verbraecken J (2019) New-generation positional therapy in patients with positional central sleep apnea. Eur Arch Otorhinolaryngol 276:1–9

Oksenberg A, Arons E, Snir D, Radwan H, Soroker N (2002) Cheyne-Stokes respiration during sleep: a possible effect of body position. Med Sci Monit 8(7):CS61–CS65

Sahlin C, Svanborg E, Stenlund H, Franklin K (2005) Cheyne-Stokes respiration and supine dependency. Eur Respir J 25(5):829–833

Soll BA, Yeo KK, Davis JW, Seto TB, Schatz IJ, Shen EN (2009) The effect of posture on Cheyne-Stokes respirations and hemodynamics in patients with heart failure. Sleep 32(11):1499–1507

Szollosi I, Roebuck T, Thompson B, Naughton MT (2006) Lateral sleeping position reduces severity of central sleep apnea/Cheyne-Stokes respiration. Sleep 29(8):1045–1051

Itasaka Y, Miyazaki S, Ishikawa K, Togawa K (2000) The influence of sleep position and obesity on sleep apnea. Psychiatry Clin Neurosci 54(3):340–341. https://doi.org/10.1046/j.1440-1819.2000.00705.x

Article   CAS   PubMed   Google Scholar  

Mo JH, Lee CH, Rhee CS, Yoon IY, Kim JW (2011) Positional dependency in Asian patients with obstructive sleep apnea and its implication for hypertension. Archiv Otolaryngol Head Neck Surg 137(8):786–790. https://doi.org/10.1001/archoto.2011.122

Teerapraipruk B, Chirakalwasan N, Simon R, Hirunwiwatkul P, Jaimchariyatam N, Desudchit T, Charakorn N, Wanlapakorn C (2012) Clinical and polysomnographic data of positional sleep apnea and its predictors. Sleep Breath 16(4):1167–1172. https://doi.org/10.1007/s11325-011-0627-5

Tanaka F, Nakano H, Sudo N, Kubo C (2009) Relationship between the body position-specific apnea-hypopnea index and subjective sleepiness. Respiration 78(2):185–190

Oksenberg A, Silverberg DS (1998) The effect of body posture on sleep-related breathing disorders: facts and therapeutic implications. Sleep Med Rev 2(3):139–162

Oksenberg A (2005) The increase of the apnoea-hypopnoea index after surgery is not always related to a real failure of the surgery. Clin Otolaryngol ENT-UK Off J Netherlands Soc Oto-Rhino-Laryngol Cervico-Facial Surg 30(5):482–483 discussion 483-486

Yalciner G, Babademez MA, Gul F (2017) Association of sleep time in supine position with apnea-hypopnea index as evidenced by successive polysomnography. Sleep Breath 21(2):289–294

Metersky ML, Castriotta RJ (1996) The effect of polysomnography on sleep position: possible implications on the diagnosis of positional obstructive sleep apnea. Respiration 63(5):283–287

Wimaleswaran H, Yo S, Buzacott H, Lim M-T, Wong A-M, Edwards B, Landry S, Thomson L, Mansfield A, Joosten S (2018) Sleeping position during laboratory polysomnography compared to habitual sleeping position at home. In: Journal of Sleep Research. Wiley, Hoboken

Logan MB, Branham GH, Eisenbeis JF, Hartse K (1996) 11: 48 AM: unattended home monitoring in the evaluation of sleep apnea: is it equal to formal polysomnograph? Otolaryngol Head Neck Surg 115(2):P156–P156

Vonk PE, Ravesloot MJL, Kasius KM, van Maanen JP, de Vries N (2019) Floppy epiglottis during drug-induced sleep endoscopy: an almost complete resolution by adopting the lateral posture. Sleep Breath 24(1):103–109. https://doi.org/10.1007/s11325-019-01847-x

Tiotiu A, Mairesse O, Hoffmann G, Todea D, Noseda A (2011) Body position and breathing abnormalities during sleep: a systematic study. Pneumologia (Bucharest, Romania) 60(4):216–221

Muraja-Murro A, Kulkas A, Hiltunen M, Kupari S, Hukkanen T, Tiihonen P, Mervaala E, Töyräs J (2013) The severity of individual obstruction events is related to increased mortality rate in severe obstructive sleep apnea. J Sleep Res 22(6):663–669

Azarbarzin A, Sands SA, Taranto-Montemurro L, Redline S, Wellman A (2019) Hypoxic burden captures sleep apnoea-specific nocturnal hypoxaemia. Eur Heart J 40:2989–2990

Leppänen T, Kulkas A, Töyräs J (2019) The hypoxic burden: also known as the desaturation severity parameter. Eur Heart J 40:2991–2993

Ravesloot MJ, de Raaff CA, van de Beek MJ, Benoist LB, Beyers J, Corso RM, Edenharter G, den Haan C, Azad JH, Ho J-PT (2019) Perioperative care of patients with obstructive sleep apnea undergoing upper airway surgery: a review and consensus recommendations. JAMA Otolaryngol Head Neck Surg

Leppänen T, Töyräs J, Muraja-Murro A, Kupari S, Tiihonen P, Mervaala E, Kulkas A (2016) Length of individual apnea events is increased by supine position and modulated by severity of obstructive sleep apnea. Sleep Disorders 2016(3):1–13

Kulkas A, Muraja-Murro A, Tiihonen P, Mervaala E, Töyräs J (2015) Morbidity and mortality risk ratios are elevated in severe supine dominant OSA: a long-term follow-up study. Sleep Breath 19(2):653–660

Saigusa H, Suzuki M, Higurashi N, Kodera K (2009) Three-dimensional morphological analyses of positional dependence in patients with obstructive sleep apnea syndrome. Anesthesiology 110(4):885–890. https://doi.org/10.1097/ALN.0b013e31819b5d57

Chang ET, Shiao GM (2008) Craniofacial abnormalities in Chinese patients with obstructive and positional sleep apnea. Sleep Med 9(4):403–410. https://doi.org/10.1016/j.sleep.2007.04.024

Walsh JH, Leigh MS, Paduch A, Maddison KJ, Armstrong JJ, Sampson DD, Hillman DR, Eastwood PR (2008) Effect of body posture on pharyngeal shape and size in adults with and without obstructive sleep apnea. Sleep 31(11):1543–1549

Soga T, Nakata S, Yasuma F, Noda A, Sugiura T, Yatsuya H, Koike Y, Ozaki N, Nakashima T (2009) Upper airway morphology in patients with obstructive sleep apnea syndrome: effects of lateral positioning. Auris Nasus Larynx 36(3):305–309. https://doi.org/10.1016/j.anl.2008.04.007

Pevernagie DA, Stanson AW, Sheedy PF 2nd, Daniels BK, Shepard JW Jr (1995) Effects of body position on the upper airway of patients with obstructive sleep apnea. Am J Respir Crit Care Med 152(1):179–185. https://doi.org/10.1164/ajrccm.152.1.7599821

Joosten SA, Sands SA, Edwards BA, Hamza K, Turton A, Lau KK, Crossett M, Berger PJ, Hamilton GS (2015) Evaluation of the role of lung volume and airway size and shape in supine-predominant obstructive sleep apnoea patients. Respirol (Carlton, Vic) 20(5):819–827. https://doi.org/10.1111/resp.12549

Malhotra A, Trinder J, Fogel R, Stanchina M, Patel SR, Schory K, Kleverlaan D, White DP (2004) Postural effects on pharyngeal protective reflex mechanisms. Sleep 27(6):1105–1112

Otsuka R, Ono T, Ishiwata Y, Kuroda T (2000) Respiratory-related genioglossus electromyographic activity in response to head rotation and changes in body position. Angle Orthodont 70(1):63–69

Joosten SA, Edwards BA, Wellman A, Turton A, Skuza EM, Berger PJ, Hamilton GS (2015) The effect of body position on physiological factors that contribute to obstructive sleep apnea. Sleep 38(9):1469–1478. https://doi.org/10.5665/sleep.4992

Joosten SA, Landry SA, Sands SA, Terrill PI, Mann D, Andara C, Skuza E, Turton A, Berger P, Hamilton GS (2017) Dynamic loop gain increases upon adopting the supine body position during sleep in patients with obstructive sleep apnoea. Respirol (Carlton, Vic) 22(8):1662–1669

Victores AJ, Hamblin J, Gilbert J, Switzer C, Takashima M (2014) Usefulness of sleep endoscopy in predicting positional obstructive sleep apnea. Otolaryngol Head Neck Surg 150(3):487–493

Lee CH, Kim DK, Kim SY, Rhee CS, Won TB (2015) Changes in site of obstruction in obstructive sleep apnea patients according to sleep position: a DISE study. Laryngoscope 125(1):248–254

Yalamanchili R, Mack WJ, Kezirian EJ (2019) Drug-induced sleep endoscopy findings in supine vs nonsupine body positions in positional and nonpositional obstructive sleep apnea. JAMA Otolaryngol Head Neck Surg 145(2):159–165

Vonk PE, van de Beek MJ, Ravesloot MJ, de Vries N (2018) Drug-induced sleep endoscopy: new insights in lateral head rotation compared to lateral head and trunk rotation in (non) positional obstructive sleep apnea patients. Laryngoscope 129(10):2430–2435

Pourhoseingholi MA, Baghestani AR, Vahedi M (2012) How to control confounding effects by statistical analysis. Gastroenterol Hepatol 5(2):79–83

Vonk PE, Rotteveel PJ, Ravesloot MJL, den Haan C, de Vries N (2019) The influence of position-dependency on surgical success in sleep apnea surgery - a systematic review. Sleep Breath 24:433–442. https://doi.org/10.1007/s11325-019-01935-y

Katsantonis GP, Miyazaki S, Walsh JK (1990) Effects of uvulopalatopharyngoplasty on sleep architecture and patterns of obstructed breathing. Laryngoscope 100(10):1068–1072

Lee CH, Kim S-W, Han K, Shin J-M, Hong S-L, Lee J-E, Rhee C-S, Kim J-W (2011) Effect of uvulopalatopharyngoplasty on positional dependency in obstructive sleep apnea. Archiv Otolaryngol Head Neck Surg 137(7):675–679

Van Maanen J, Ravesloot M, Witte B, Grijseels M, De Vries N (2012) Exploration of the relationship between sleep position and isolated tongue base or multilevel surgery in obstructive sleep apnea. Eur Arch Otorhinolaryngol 269(9):2129–2136

Lee YC, Eun YG, Shin SY, Kim SW (2014) Change in position dependency in non-responders after multilevel surgery for obstructive sleep apnea: analysis of polysomnographic parameters. Eur Arch Otorhinolaryngol 271(5):1081–1085

Li H-Y, Cheng W-N, Chuang L-P, Fang T-J, Hsin L-J, Kang C-J, Lee L-A (2013) Positional dependency and surgical success of relocation pharyngoplasty among patients with severe obstructive sleep apnea. Otolaryngol Head Neck Surg 149(3):506–512

van Maanen JP, Witte BI, de Vries N (2014) Theoretical approach towards increasing effectiveness of palatal surgery in obstructive sleep apnea: role for concomitant positional therapy? Sleep Breath 18(2):341–349. https://doi.org/10.1007/s11325-013-0891-7

Kastoer C, Benoist L, Dieltjens M, Torensma B, de Vries L, Vonk P, Ravesloot M, de Vries N (2018) Comparison of upper airway collapse patterns and its clinical significance: drug-induced sleep endoscopy in patients without obstructive sleep apnea, positional and non-positional obstructive sleep apnea. Sleep Breath 22(4):939–948

Benoist LB, de Ruiter MH, de Lange J, de Vries N (2015) Residual POSA after maxillomandibular advancement in patients with severe OSA. In: Positional therapy in obstructive sleep apnea. Springer, Berlin, pp 321–329

Vonk PE, Rotteveel PJ, Ravesloot MJ, Ho J-PT, de Lange J, de Vries N (2020) The influence of position dependency on surgical success in patients with obstructive sleep apnea undergoing maxillomandibular advancement. J Clin Sleep Med 16(1):73–80

Oksenberg A, Goizman V, Eitan E, Nasser K, Gadoth N, Leppänen T (2019) Obstructive sleep apnea: do positional patients become nonpositional patients with time? Laryngoscope 130(9):2263–2268

Benoist L, Verhagen M, Torensma B, van Maanen J, De Vries N (2017) Positional therapy in patients with residual positional obstructive sleep apnea after upper airway surgery. Sleep Breath 21(2):279–288

Steffen A, Hartmann JT, König IR, Ravesloot MJ, Hofauer B, Heiser C (2018) Evaluation of body position in upper airway stimulation for obstructive sleep apnea—is continuous voltage sufficient enough? Sleep Breath 22(4):1207–1212

Lee CH, Shin H-W, Han DH, Mo J-H, Yoon I-Y, Chung S, Choi HG, Kim J-W (2009) The implication of sleep position in the evaluation of surgical outcomes in obstructive sleep apnea. Otolaryngol Head Neck Surg 140(4):531–535

Dieltjens M, Braem MJ, Van de Heyning PH, Wouters K, Vanderveken OM (2014) Prevalence and clinical significance of supine-dependent obstructive sleep apnea in patients using oral appliance therapy. J Clin Sleep Med 10(9):959–964. https://doi.org/10.5664/jcsm.4024

Sutherland K, Takaya H, Qian J, Petocz P, Ng AT, Cistulli PA (2015) Oral appliance treatment response and polysomnographic phenotypes of obstructive sleep apnea. J Clin Sleep Med 11(08):861–868

Chung JW, Enciso R, Levendowski DJ, Morgan TD, Westbrook PR, Clark GT (2010) Treatment outcomes of mandibular advancement devices in positional and nonpositional OSA patients. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 109(5):724–731. https://doi.org/10.1016/j.tripleo.2009.11.031

Lee CH, Jung HJ, Lee WH, Rhee CS, Yoon IY, Yun PY, Kim JW (2012) The effect of positional dependency on outcomes of treatment with a mandibular advancement device. Archiv Otolaryngol Head Neck Surg 138(5):479–483. https://doi.org/10.1001/archoto.2012.452

Marklund M, Stenlund H, Franklin KA (2004) Mandibular advancement devices in 630 men and women with obstructive sleep apnea and snoring: tolerability and predictors of treatment success. Chest 125(4):1270–1278

Morong S, Benoist LB, Ravesloot MJ, Laman DM, de Vries N (2014) The effect of weight loss on OSA severity and position dependence in the bariatric population. Sleep Breath 18(4):851–856. https://doi.org/10.1007/s11325-014-0955-3

Oksenberg A, Dynia A, Nasser K, Gadoth N (2012) Obstructive sleep apnoea in adults: body postures and weight changes interactions. J Sleep Res 21(4):402–409. https://doi.org/10.1111/j.1365-2869.2011.00988.x

Joosten SA, Khoo JK, Edwards BA, Landry SA, Naughton MT, Dixon JB, Hamilton GS (2017) Improvement in obstructive sleep apnea with weight loss is dependent on body position during sleep. Sleep 40(5):zsx047

Heinzer R, Vat S, Marques-Vidal P, Marti-Soler H, Andries D, Tobback N, Mooser V, Preisig M, Malhotra A, Waeber G (2015) Prevalence of sleep-disordered breathing in the general population: the HypnoLaus study. Lancet Respir Med 3(4):310–318

CAS   PubMed   PubMed Central   Google Scholar  

Ravesloot MJ, de Vries N (2011) Reliable calculation of the efficacy of non-surgical and surgical treatment of obstructive sleep apnea revisited. Sleep 34(1):105–110

Ravesloot MJ, de Vries N, Stuck BA (2014) Treatment adherence should be taken into account when reporting treatment outcomes in obstructive sleep apnea. Laryngoscope 124(1):344–345. https://doi.org/10.1002/lary.24302

Stuck BA, Leitzbach S, Maurer JT (2012) Effects of continuous positive airway pressure on apnea–hypopnea index in obstructive sleep apnea based on long-term compliance. Sleep Breath 16(2):467–471

Grote L, Hedner J, Grunstein R, Kraiczi H (2000) Therapy with nCPAP: incomplete elimination of Sleep Related Breathing Disorder. Eur Respir J 16(5):921–927

Vanderveken OM, Dieltjens M, Wouters K, De Backer WA, Van de Heyning PH, Braem MJ (2012) Objective measurement of compliance during oral appliance therapy for sleep-disordered breathing. Thorax 68(1):91–96

Bianchi MT, Alameddine Y, Mojica J (2014) Apnea burden: efficacy versus effectiveness in patients using positive airway pressure. Sleep Med 15(12):1579–1581

Boyd SB, Walters AS (2013) Effectiveness of treatment apnea-hypopnea index: a mathematical estimate of the true apnea-hypopnea index in the home setting. J Oral Maxillofac Surg 71(2):351–357. https://doi.org/10.1016/j.joms.2012.05.009

Cartwright RD, Diaz F, Lloyd S (1991) The effects of sleep posture and sleep stage on apnea frequency. Sleep 14(4):351–353

van Maanen JP, Richard W, Van Kesteren ER, Ravesloot MJ, Laman DM, Hilgevoord AA, de Vries N (2012) Evaluation of a new simple treatment for positional sleep apnoea patients. J Sleep Res 21(3):322–329. https://doi.org/10.1111/j.1365-2869.2011.00974.x

Ravesloot MJ, van Maanen JP, Dun L, de Vries N (2013) The undervalued potential of positional therapy in position-dependent snoring and obstructive sleep apnea-a review of the literature. Sleep Breath 17(1):39–49. https://doi.org/10.1007/s11325-012-0683-5

Levendowski DJ, Seagraves S, Popovic D, Westbrook PR (2014) Assessment of a neck-based treatment and monitoring device for positional obstructive sleep apnea. J Clin Sleep Med 10(8):863–871. https://doi.org/10.5664/jcsm.3956

Ravesloot MJ, White D, Heinzer R, Oksenberg A, Pépin J-L (2017) Efficacy of the new generation of devices for positional therapy for patients with positional obstructive sleep apnea: a systematic review of the literature and meta-analysis. J Clin Sleep Med 13(06):813–824

Sutherland K, Phillips CL, Cistulli PA (2015) Efficacy versus effectiveness in the treatment of obstructive sleep apnea: CPAP and oral appliances. J Dental Sleep Med 2(4):175–181

Eijsvogel MM, Ubbink R, Dekker J, Oppersma E, de Jongh FH, van der Palen J, Brusse-Keizer MG (2015) Sleep position trainer versus tennis ball technique in positional obstructive sleep apnea syndrome. J Clin Sleep Med 11(2):139–147. https://doi.org/10.5664/jcsm.4460

Ravesloot MJ, Benoist L, van Maanen P, de Vries N (2017) Novel positional devices for the treatment of positional obstructive sleep apnea, and how this relates to sleep surgery. In: Sleep-related breathing disorders, vol 80. Karger Publishers, Berlin, pp 28–36

Bignold JJ, Deans-Costi G, Goldsworthy MR, Robertson CA, McEvoy D, Catcheside PG, Mercer JD (2009) Poor long-term patient compliance with the tennis ball technique for treating positional obstructive sleep apnea. J Clin Sleep Med 5(5):428–430

Oksenberg A, Silverberg D, Offenbach D, Arons E (2006) Positional therapy for obstructive sleep apnea patients: a 6-month follow-up study. Laryngoscope 116(11):1995–2000. https://doi.org/10.1097/01.mlg.0000237674.66716.a7

Heinzer RC, Pellaton C, Rey V, Rossetti AO, Lecciso G, Haba-Rubio J, Tafti M, Lavigne G (2012) Positional therapy for obstructive sleep apnea: an objective measurement of patients’ usage and efficacy at home. Sleep Med 13(4):425–428. https://doi.org/10.1016/j.sleep.2011.11.004

van Maanen JP, Meester KA, Dun LN, Koutsourelakis I, Witte BI, Laman DM, Hilgevoord AA, de Vries N (2013) The sleep position trainer: a new treatment for positional obstructive sleep apnoea. Sleep Breath 17(2):771–779. https://doi.org/10.1007/s11325-012-0764-5

Laub RR, Tønnesen P, Jennum PJ (2017) A Sleep Position Trainer for positional sleep apnea: a randomized, controlled trial. J Sleep Res 26(5):641–650

de Ruiter MH, Benoist LB, De Vries N, de Lange J (2018) Durability of treatment effects of the Sleep Position Trainer versus oral appliance therapy in positional OSA: 12-month follow-up of a randomized controlled trial. Sleep Breath 22(2):441–450

Beyers J, Vanderveken OM, Kastoer C, Boudewyns A, De Volder I, Van Gastel A, Verbraecken JA, De Backer WA, Braem MJ, Van de Heyning PH, Dieltjens M (2019) Treatment of sleep-disordered breathing with positional therapy: long-term results. Sleep Breath 23(4):1141–1149. https://doi.org/10.1007/s11325-019-01792-9

van Maanen JP, de Vries N (2014) Long-term effectiveness and compliance of positional therapy with the sleep position trainer in the treatment of positional obstructive sleep apnea syndrome. Sleep 37(7):1209–1215. https://doi.org/10.5665/sleep.3840

Berry RB, Uhles ML, Abaluck BK, Winslow DH, Schweitzer PK, Gaskins RA Jr, Doekel RC Jr, Emsellem HA (2019) NightBalance sleep position treatment device versus auto-adjusting positive airway pressure for treatment of positional obstructive sleep apnea. J Clin Sleep Med 15(7):947–956

Levendowski DJ, Oksenberg A, Vicini C, Penzel T, Levi M, Westbrook PR (2018) A systematic comparison of factors that could impact treatment recommendations for patients with positional obstructive sleep apnea (POSA). Sleep Med 50:145–151

Berry RB, Brooks R, Gamaldo CE, Harding SM, Marcus C, Vaughn BV (2012) The AASM manual for the scoring of sleep and associated events. In: Rules, Terminology and Technical Specifications, vol 176. American Academy of Sleep Medicine, Darien, p 2012

Tate A, Walsh J, Kurup V, Shenoy B, Mann D, Freakley C, Eastwood P, Terrill P (2020) An emerging technology for the identification and characterization of postural-dependent obstructive sleep apnea. J Clin Sleep Med 16(2):309–318

Download references

Author information

Authors and affiliations.

Department of Otorhinolaryngology - Head and Neck surgery, OLVG, location West, Jan Tooropstraat 164, Jan Tooropstraat 164, 1061 AE, Amsterdam, the Netherlands

M. J. L. Ravesloot & N. de Vries

Department of Otorhinolaryngology - Head and Neck surgery, Amsterdam UMC, location AMC, Amsterdam, the Netherlands

Department of Otorhinolaryngology, Head and Neck Surgery, Sleep Disorders Center, University Hospital Mannheim, Mannheim, Germany

J.T. Maurer

Sleep Disorders Unit, Loewenstein Hospital-Rehabilitation Center, Raanana, Israel

A. Oksenberg

Department of Oral Kinesiology, ACTA, MOVE Research Institute Amsterdam, University of Amsterdam and VU University Amsterdam, Amsterdam, the Netherlands

N. de Vries

Department of Otorhinolaryngology - Head and Neck Surgery, Antwerp University Hospital (UZA), Antwerp, Belgium

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to M. J. L. Ravesloot .

Ethics declarations

Conflict of interest.

Nico de Vries is a member of the Medical Advisory Board of NightBalance. He is also an investigator of Inspire and Jazz Pharmaceutics, consultant of Philips, Olympus, and the AE Mann Foundation.

The authors have no other funding, financial relationships, or conflicts of interest to disclose.

Code availability

Additional information, publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Reprints and permissions

About this article

Ravesloot, M.J.L., Vonk, P.E., Maurer, J. et al. Standardized framework to report on the role of sleeping position in sleep apnea patients. Sleep Breath 25 , 1717–1728 (2021). https://doi.org/10.1007/s11325-020-02255-2

Download citation

Received : 23 June 2020

Revised : 09 November 2020

Accepted : 13 November 2020

Published : 11 January 2021

Issue Date : December 2021

DOI : https://doi.org/10.1007/s11325-020-02255-2

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• “Sleep Apnea Obstructive/therapy” [Mesh]
• “Supine Position” [MeSh]
• Positional therapy
• Obstructive sleep apnea
• Sleeping position
• Find a journal
• Publish with us
• Track your research

• Open access
• Published: 08 September 2020

The anti-snoring bed - a pilot study

• Elisabeth Wilhelm 1 ,
• Francesco Crivelli 1 , 2 ,
• Nicolas Gerig 1 , 3 ,
• Malcolm Kohler 4 &
• Robert Riener 1 , 5  

Sleep Science and Practice volume  4 , Article number:  14 ( 2020 ) Cite this article

11k Accesses

2 Citations

19 Altmetric

Metrics details

Avoiding supine position can reduce snoring in most habitual snorers. However, devices that restrict the sleeping position cause discomfort or disrupt sleep resulting in low compliance. Therefore, mechanisms, which lift the trunk of the user without disturbing sleep, have been proposed. We present the first study, which investigates whether individual interventions provided by beds with lifting mechanisms are able to stop snoring (success rate) and whether they reduce the snoring index (number of total snores divided by total time in bed) using a repeated measures design. In addition, we investigated whether the intervention is interfering with the subjective sleep quality.

Twenty-two subjects were observed for four nights (adaptation, baseline, and two intervention nights). During intervention nights, the bed lifted the trunk of the user in closed-loop manner. Subjects were divided in three groups ( non-snorer , snorer one , and snorer two ). Non-snorers were lifted by the bed at random time points during the night. In group snorer one , a stepwise increase of the bed inclination was compared with going directly to a randomly selected angle. In group snorer two , the influence of a small inclination angle (10 ∘ ) and a big inclination angle (20 ∘ ) was compared.

Snoring was stopped successfully in 22% (small angle) and 67% (big angle) of the interventions. This did not lead to a significant reduction in the snoring index. The subjective sleep quality was not reduced by the intervention.

The anti-snoring bed is able to stop individual episodes of habitual snoring without reducing the subjective sleep quality.

Trial Registration

https://clinicaltrials.gov , no. NCT04053738, registered 12 August 2019 - Retrospectively registered, https://clinicaltrials.gov/ct2/show/NCT04053738 .

Introduction

Habitual snoring is a widespread sleep problem ( Deary et al. 2014 ), which does not only affect the health of the snorer ( Endeshaw et al. 2013 ) but also the quality of life of the bed partner ( Beninati et al. 1999 ). Intense snorers snore up to 245 times/hour ( Cathcart et al. 2010 ). Snoring often occurs when the muscle tone drops while the snorer is lying in supine position. In this position, gravitational forces pull the soft tissue surrounding the upper airways in dorsal direction, thereby narrowing the airways. This causes audible flow turbulences in the upper airways ( Gleadhill et al. 1991 ).

Therapeutic interventions force snorers to avoid supine position. The easiest way to implement positional therapy is to attach an obstacle, to the back of the snorer ( Berger et al. 1997 ). More sophisticated devices include vibration ( van Maanen and de Vries 2014 ) or sound ( Cartwright et al. 1985 ) alarms raised when the snorer sleeps in supine position, or cushions, which push the head of the snorer sideward ( Zhang et al. 2013 ). However, the discomfort needed to force the snorer to avoid supine position results in disrupted sleep and low user compliance ( Bignold et al. 2009 ).

An alternative approach is to elevate the trunk of the user with cushions. McEvoy et al. reported that sleeping in seated position (trunk elevation of 60 ∘ ) has positive influences on the severity of obstructive sleep apnea ( McEvoy et al. 1986 ). Recently it has been suggested to use adjustable bed frames to change the position of the sleeping snorer ( Van Der Loos et al. 2003 ). Actuated beds such as the one used within this paper cost about 7000 Swiss francs and are therefore more expensive than conventional approaches for providing positional therapy. Commercially available versions such as the Partner Snore TM Technology (Sleep Number Corporation, Minneapolis, USA) elevate the trunk of the sleeping snorer if the bed partner indicates the presence of snoring via a push button ( Sleep Number Corporation ). Despite the fact that these beds are commercially available, little is known about the effect of such devices.

In this paper, we present the first systematic pilot study, which investigates the influence of automated trunk elevations on snorers. In absence of any evidence on the effect of different intervention parameters, we decided to conduct a pilot study, which investigates multiple options with three small groups. This pilot study aims on identifying the range of promising intervention options for future investigations. In particular, we were interested, which elevation angles would most likely reduce habitual snoring. Furthermore, we investigated how the bed influences the subjective sleep quality of the user.

Materials and methods

The mattress shape was adapted using a custom made intelligent anti-snoring bed (Fig.  1 ), which is able to detect snoring sound and change the position of the user whenever snoring occurs ( Crivelli et al. 2017 ; Wilhelm et al. 2017 ). The bed is triggered depending on the snoring index (SI) defined as number of snores detected within a given time frame. This time frame was set to one minute. The threshold that triggered bed motion was ten snores per minute for the first part of the study. After the interim analysis, this threshold was lowered to four snores per minute.

Measurement setup. a Postural interventions were provided using an automated bed, which can elevate the trunk of the sleeping user. A custom-made built-in microphone was used to trigger trunk elevation. In addition, a reference microphone and a IR-camera were used to monitor the user. b The inclination angles provided by the bed were 10 ∘ (P10), 15 ∘ (P15), and 20 ∘ (P20) provided at the location where the users hip joint is to be expected. At the beginning of the night on during episodes of silence the bed was set to neutral postion (P0)

Twenty-two participants (5 female, age 29.7 years SD: 16.0) participated in the study (Table  1 ). Participants were not medication free. Alcohol and caffeine consumption was not restricted. Exclusion criteria were pregnancy, previously diagnosed sleep-related breathing disorders, chronic lower back pain, heart insufficiency that might impede sleeping in supine position, and inability to follow the procedures of the study. All subjects received a monetary compensation (CHF 25 per night) to cover travel costs. Daytime sleepiness was recorded using the Epworth Sleepiness Scale (ESS). Ten participants reached an ESS-score above 10, which indicates excessive daytime sleepiness ( Johns 1991 ). Apnea hypopnea index (AHI) and oxygen desaturation index (ODI) were measured using a home-monitoring device (Apnea Link, purchased from ResMed). The scoring was done automatically by the ApneaLink software (purchased from ResMed). Two participants had an AHI and an ODI bigger than 5 during the adaptation night. One of these two participants also scored more than 10 points in the ESS. Therefore, we cannot exclude that these participants suffer from an undiagnosed sleep-related breathing disorder. In addition, participants were asked to report any breathing related co-morbidity. One participant reported being allergic to cats and having frequent breathing problems after smoking for about 6 years. Another participant reported being affected by allergic rhinitis (hay fever). Four participants reported experiencing symptoms of a cold on one or more days on which they participated in the study.

Participants were divided into snorers and non-snorers based on the amount of snoring recorded with a home-monitoring device (Apnea Link, purchased from ResMed) during the first night. The threshold was set to 50 snores per night. This value was chosen with respect to the normal breath rate of a human subject (12 -20 breath/minute), someone who snores 50 times per night snores between 2.5 and 5 min in total. Interventions by the bed are only triggered after 1 minute of continuous snoring. Therefore, the chance to see interventions in snorers who snore less than 50 times per night would be very low. Furthermore, by excluding subjects with a snoring episode of less than 5 min we wanted to avoid overestimation of an overestimation of positive interventions. Positive interventions were defined as bad movements (which are triggered after 1 minute of snoring) that were able to stop snoring within 3 minutes time. Participants were recruited in two phases. The group snorer one was recruited between February and July 2017. The group snorer two was recruited from February to July 2018. All non-snorers recruited in both recruitment periods received the same interventions. Parameters for snorers were optimized after completion of the recruitment period.

One subject withdrew from the study. Four subjects did not receive the intervention in at least one night of the study due to technical problems. Three snorers did not trigger the bed. These datasets were excluded from the evaluation. Some subjects only triggered the bed in one of the experimental conditions. Nights in which the bed was not triggered were treated as missing data points.

Study protocol

All subjects spent four consecutive nights in the lab. Daytime activities were not restricted. Subjects went to bed at their habitual bedtime. The first night served as adaptation night. Night two to four were used as experimental nights. These nights were split into one baseline night and two nights with interventions. The order of the nights was randomized.

Within this study, we used four different elevation angles (0, 10, 15, and 20 degree referred to as P0, P10, P15, P20, respectively). The bed moved with an angular velocity of 1.5 ∘ /s. As depicted in Fig.  2 the three groups got different interventions.

Interventions provided by the bed. Participants of the groups snorer one and non-snorer were provided with a stepwise intervention and a random angle intervention. During stepwise intervention, the bed was elevated upon the detection of snoring. If snoring stopped within 5 minutes, the bed went back to P0. Otherwise, the inclination was increased to the next step. During the random angle intervention, the bed went to one of the three inclination angles upon detection of snoring and waited in this position until snoring stopped. The group snorer two got two different interventions. During the small angle intervention, the bed moved to P10 upon detection of snoring and back to neutral when snoring stopped. During the big angle intervention, the bed went to P20

Within the groups snorer one and non-snorer we wanted to investigate whether a stepwise increasing inclination angle ( stepwise condition) is less disturbing than a sudden change of position ( random angle condition) (Fig.  2 ). Since Non-snorers would not trigger the bed with snoring sounds, interventions were provided to them at random time points throughout the night.

We compared the lowest angle (10 ∘ , P10) with the highest inclination angle (20 ∘ , P20) used within the group snorer two . Nights during which the bed only went to P10 are called small angle . The other nights are referred to as big angle .

Goal of the study

Within this study, we investigated, whether individual interventions of the bed are able to stop snoring episodes. In addition, we were interested, whether the overall amount of snoring would be reduced in intervention nights compared to baseline. Furthermore, we investigated whether the bed influences the subjective sleep quality.

Snoring activity was monitored using a portable home-monitoring device (Apnea Link plus, purchased from ResMed, Switzerland). Since audio based detection has been reported to be the most reliable way of snoring detection ( Arnardottir et al. 2016 ), we also used a sound level meter (XL2 Audio and acoustic Analyzer, purchased from NTi Audio, Lichtenstein). Manual visual-audio scoring of the data collected with the noise level meter was performed using the XL2 data explorer.

The success rate of the individual intervention was derived by dividing the number of positive interventions through the total number of bed movements.

To investigate the effect on the total amount of snoring observed within one night, we calculated the SI using the lables of the software of the Apnea Link and the lables of the manual scoring. The SI was calculated by dividing the number of snores ( S ) through the recording time ( t record ):

The sleeping position of the user was manually scored based on an infrared video. We distinguished between supine, prone, left, or right position and movement episodes.

Subjective Sleep Quality was assessed using the Groningen Sleep Quality Scale (GSQS) ( Meijman et al. 1988 ). In addition, subjects were asked to indicate whether they think that an intervention occurred during the night, whether they woke up due to an intervention, and whether the intervention was disturbing their sleep. On the morning after the last experimental night, subjects were also asked to indicate which condition they preferred. The questionnaires were filled out within the first 10 minutes after wake up.

Statistical analysis

The data was analysed using Matlab r2018b (MathWorks, USA). Data was tested for normality using the Kolmogorow-Smirnow-Test. Since this test indicated a non-normal distribution, we used a non-parametric method. The Friedman-Test, which would be the non-parametric alternative for an ANOVA in repeated measures design, could not be used, since it does not accept missing values. Alternative tests, which do accept missing values, require a bigger sample size. Therefore, we used the Kruskal-Wallis Test to compare the values within each group. All results are reported as mean and range. The significance level was set to p =0.05.

We decided to look at each group individually during the analysis, because the number of bed movements differed in between groups.

Number of interventions triggered by different groups

Non-snorers received 12 bed movements in both random angle and stepwise condition. The group Snorer one triggered the bed on average 1.5 (range: 0, 4) times and 2 (range: 0, 7) times in stepwise and random angle condition, respectively. In stepwise condition , all but one snorer stopped snoring when the bed reached P10. The other snorer stopped snoring when the bed was in P15. Averaged over all nights snorers of the group snorer one only encountered 6.3 (range: 0, 13) bed movements. This is only half of what the control group experienced. Therefore, we decided to investigate at each group individually.

Participants of the group snorer two triggered the bed on average 3 (range: 0, 11) and 2 (range: 0, 6) times in the small and big angle condition, respectively.

Effects of the postural intervention on snoring

In the group snorer one 60.0 % of the stepwise interventions and 57.0 % of the random angle interventions were successful. As depicted in Fig.  3 the bed reduced the audio-based SI slightly in snorer one during the random angle condition (n.s.). In the data recorded with the home-monitoring device there was no reduction of snoring in intervention nights.

Effect of the automated bed on snoring. The average SI was once derived from manually scored audio-recording and once from the nasal cannula of the home-monitoring device. As depicted the bed was able to stop snoring with a success rate of 22.2 % to 66.6 % in different test conditions. However, the successful interventions did not lead to a reduction of the average SI which describes the total amount of snores divided by the time spent in bed

In the group snorer two , small angle was provided 18 times with a success rate of 22.2 %. The intervention big angle was triggered 12 times with a success rate of 66.6 %. There was no statistical significant difference between the audio-based SI recorded during the different nights. The same applied for the SI measured with the home-monitoring device. However, both outcome measures showed a tendency towards a lower SI in the big angle condition (see Fig.  3 ).

Effects of the postural intervention on subjective sleep quality

As depicted in Fig.  4 the stepwise intervention seemed to decrease subjective sleep quality measured using the GSQS in the groups non-snorer and in snorer one (n.s.). Participants of both groups noticed in which nights interventions occurred. This result was significant in snorer one ( \(p= 0.0498, \tilde {\chi }^{2}= 6.0000\) , with a mean rank of 7.5, 3.0, and 3.0 for baseline, stepwise , and random angle respectively) but not in non-snorer . The questions whether subjects woke up due to an intervention and whether the bed did disturb the participants did not give significant results. Two out of four snorers and two out of six non-snorers did not indicate which condition they preferred. One out of six non-snorer chose the stepwise condition and one the random angle condition. All other participants indicated that they preferred the baseline condition.

Influence on subjective sleep quality. GSQS scores reach from 0 to 14 with 0 indicating good sleep and 14 indicating bad sleep. As depicted non-snorers slept worse during intervention nights. In group snorer one sleep quality improved slightly during the random angle condition. The group snorer two had a tendency to report better sleep quality in both intervention conditions. None of these results was significant

In the group snorer two , we investigated the influence of small and big inclination angles. As depicted in Fig.  4 . there was no difference in GSQS. Subjects of snorer two noticed in which nights interventions occurred ( \( p= 0.0481, \tilde {\chi }^{2}= 6.0701\) ) with a mean rank of 13.67, 8.17, and 6.67 for baseline, small angle , and big angle , respectively. Furthermore, they reported that they had been awake due to the intervention ( \( p= 0.0439, \tilde {\chi }^{2}= 6.2503\) , mean rank of 5.3, 10.25, and 12.92 for baseline, small angle , and big angle , respectively). However, the question whether the intervention disturbed their sleep was not giving significant results. Three out of six participants of the group snorer two reported that the big angle condition was their favourite condition, two opted for baseline, and one for small angle condition.

Within our study, snoring stopped in 60 %, 57 %, or 67 %, while a intervention in stepwise , random angle , or big angle , was provided, respectively. However, this did not lead to a significant reduction of the average SI. Eventually, the natural fluctuation of snoring covered the effect of the individual interventions. Within the same subject snoring frequency and the duration of snoring episodes can vary by 22 % and 33 %, respectively, in consecutive nights ( Cathcart et al. 2010 ). Therefore, long-term studies might be needed to evaluate the effect of anti-snoring beds. While the natural variation of habitual snoring makes it hard to evaluate scientific studies, it also shows how important it is to provide interventions in closed-loop manner. The closed-loop approach allows the user to sleep in the sleeping position he prefers and will only intervene if snoring occurs. Therefore, it is much less restrictive than commonly used approaches.

In addition, our study had some limitations. In some nights, the nasal cannula was displaced due to movement. In these cases, the SI was only calculated for periods with valid measurements. Since the snoring activity varies throughout the different sleep stages this reduces the validity of this measure. The audio-based evaluation has been reported to be more reliable. However, due to the absence of a clear definition of snoring sound, this evaluation is influenced by the subjective experience of the expert scorer ( Rohrmeier et al. 2014 ).

Furthermore, there are multiple causes for habitual snoring. We recruited snorers based on self-report. Therefore, we had a mixed population. So far, it has been suggested that anti-snoring beds can only be effective for position dependent snorers. One reason why we did not see significant changes of the SI could be that both our snorer groups did also contain snorers who were not position dependent.

Another limitation of our study is, that in non-snorers interventions were provided regardless of the sleeping position. Position dependent snorers would be expected to trigger the bed while lying in supine position. In the non-snorers in our study only 24% of the upward movements of the stepwise condition and 47% of the upward movements of the random angle condition occurred while the participants were lying in supine position. This might be due to the fact that only three out of the six participants of this groups spent more than 50% of the baseline night in supine position. By recruiting non-snorers who are mainly sleeping in supine position, one could investigate the effect of the intervention in this specific sleeping position.

In addition, the snoring detection algorithm of the anti-snoring bed did not recognize all types of snoring. Therefore, some snorers did not trigger the bed. Eight times the bed reacted to environmental noise. This implies that more robust snoring detection algorithms could decrease sleep disturbance caused by anti-snoring beds. Alternatively, the interventions provided by the bed could be triggered manually by an expert scorer. However, this solution would be very cost intensive and could therefore be only used in scientific studies and not as a long-term treatment for habitual snoring.

With respect to the effect of anti-snoring beds on subjective sleep quality, we saw a slight tendency towards an increased subjective sleep quality in snorers in conditions in which the bed moved to a target position and back. In the stepwise condition, the subjective sleep quality decreased. This might be because the bed moved more frequently in this condition. The effect of an intervention on the objective sleep quality can only be investigated using polysomnography. However, polysomnography is highly obtrusive and would reduce the subjective sleep quality of the participants. As reduced comfort is the most common reason for aborting conventional therapy approaches, our approach would only be a real alternative if the subjective sleep quality is at least similar to baseline condition. Therefore, we decided not to investigate objective sleep quality measures and did instead focus on the subjective sleep quality in our pilot study. In future studies polysomnography should be included to investigate whether the motion of the bed causes arousals or changes the sleep architecture.

In our Anti-snoring bed study, snoring stopped more frequently when interventions of the type big angle (20 ∘ ) were provided compared to interventions of the type small angle (10 ∘ ). This suggests that larger inclination angles might be more efficient. Furthermore, conditions with less frequent bed movements that cover a big range of motion seem to have less effect on subjective sleep quality than conditions with multiple smaller bed movements. The interventions provided by the bed did not lead to a significant reduction in subjective sleep. Further studies are needed to investigate whether Anti-snoring beds are a valuable alternative to conventional positional therapy.

Availability of data and materials

Video and audio data generated during the current study are not publicly available, because this data could not be anonymized. Tables containing the GSQS, the snoring indices, the manual scoring of the audio data of the snorers, the manual scoring of the video data of the snorers, and the scores obtained with the home-monitoring device are available on reasonable request in the ETH research collection repository under https://www.research-collection.ethz.ch/handle/20.500.11850/396977 .

Abbreviations

Body Mass Index

Swiss Franc

Epworth Sleepiness Scale

not significant

Elevation angle 0 ∘

Elevation angle 10 ∘

Elevation angle 15 ∘

Elevation angle 20 ∘

Number of snores

Standard deviation

Snoring index

Recording time

Groningen Sleep Quality Scale

Apnea Hypopnea Index

Oxygen Desaturation Index

Arnardottir, ES, Isleifsson B, Agustsson JS, Sigurdsson GA, Sigurgunnarsdottir MO, Sigurdarson GT, Saevarsson G, Sveinbjarnarson AT, Hoskuldsson S, Gislason T. How to measure snoring? a comparison of the microphone, cannula and piezoelectric sensor. J Sleep Res. 2016; 25(2):158–68.

Article   Google Scholar  

Beninati, W, Harris CD, Herold DL, Shepard Jr JW. The Effect of Snoring and Obstructive Sleep Apnea on the Sleep Quality of Bed Partners. Mayo Clinic Proceedings. 1999; 74(10):955–8. https://doi.org/10.4065/74.10.955. http://www.sciencedirect.com/science/article/pii/S0025619611639918 .

Berger, M, Oksenberg A, Silverberg D, Arons E, Radwan H, Iaina A. Avoiding the supine position during sleep lowers 24 h blood pressure in obstructive sleep apnea (osa) patients. J Hum Genet. 1997; 11(10):657–64.

CAS   Google Scholar  

Bignold, JJ, Deans-Costi G, Goldsworthy MR, Robertson CA, McEvoy D, Catcheside PG, Mercer JD. Poor long-term patient compliance with the tennis ball technique for treating positional obstructive sleep apnea. J Clin Sleep Med. 2009; 5(05):428–30.

Cartwright, RD, Lloyd S, Lilie J, Kravitz H. Sleep position training as treatment for sleep apnea syndrome: a preliminary study. Sleep. 1985; 8(2):87–94.

Article   CAS   Google Scholar  

Cathcart, R, Hamilton D, Drinnan M, Gibson G, Wilson J. Night-to-night variation in snoring sound severity: one night studies are not reliable. Clin Otolaryngol. 2010; 35(3):198–203.

Crivelli, F, Wilhelm E, van Sluijs R, Riener R. Actuated bed for a closed loop anti-snoring therapy. In: 2017 International Conference on Rehabilitation Robotics (ICORR). London: IEEE: 2017. p. 823–7.

Google Scholar  

Deary, V, Ellis JG, Wilson JA, Coulter C, Barclay NL. Simple snoring: not quite so simple after all?Sleep Med Rev. 2014; 18(6):453–62.

Endeshaw, Y, Rice TB, Schwartz AV, Stone KL, Manini TM, Satterfield S, Cummings S, Harris T, Pahor M, Study HA. Snoring, daytime sleepiness, and incident cardiovascular disease in the health, aging, and body composition study. Sleep. 2013; 36(11):1737–45.

Gleadhill, IC, Schwartz AR, Schubert N, Wise RA, Permutt S, Smith PL. Upper airway collapsibility in snorers and in patients with obstructive hypopnea and apnea. Am Rev Respir Dis. 1991; 143(6):1300–3.

Johns, MW. A new method for measuring daytime sleepiness: the epworth sleepiness scale. Sleep. 1991; 14(6):540–5.

McEvoy, RD, Sharp DJ, Thornton AT. The effects of posture on obstructive sleep apnea. Am Rev Respir Dis. 1986; 133(4):662–6.

CAS   PubMed   Google Scholar  

Meijman, T, de Vries-Griever A, De Vries G, Kampman R. The evaluation of the groningen sleep quality scale. Groningen: Heymans Bull (HB 88-13-EX). 1988; 2006.

Rohrmeier, C, Herzog M, Ettl T, Kuehnel TS. Distinguishing snoring sounds from breath sounds: a straightforward matter?Sleep Breathing. 2014; 18(1):169–76.

Sleep Number Corporation. FlexFit ™ smart adjustable bases. https://www.sleepnumber.com/sn/en/c/bed-bases . Accessed 17 Feb 2019.

Van Der Loos, HM, Ullrich N, Kobayashi H. Development of sensate and robotic bed technologies for vital signs monitoring and sleep quality improvement. Auton Robot. 2003; 15(1):67–79.

van Maanen, JP, de Vries N. Long-term effectiveness and compliance of positional therapy with the sleep position trainer in the treatment of positional obstructive sleep apnea syndrome. Sleep. 2014; 37(7):1209–15.

Wilhelm, E, Crivelli F, Markic S, Kohler M, Pugliese F, Riener R. Ein intelligentes bett zur verminderung von schnarchen. Somnologie. 2017; 21(S2):52.

Zhang, J, Zhang Q, Wang Y, Qiu C. A real-time auto-adjustable smart pillow system for sleep apnea detection and treatment. In: 2013 ACM/IEEE International Conference on Information Processing in Sensor Networks (IPSN). Philadelphia: IEEE: 2013. p. 179–90.

Download references

Acknowledgements

We thank François Pugliese from Elite SA for his support throughout the project. In addition, we thank Michael Herold-Nadig and Marco Bader from ETH Zurich for their technical support, their advice and contribution during the design and the development of the devices. Furthermore, we thank Rachel van Sluijs, Antonino Crivello, Quincy Rondei, and Alexander Breuss for their support during the measurements. We also thank Jana Petr for her support during sound file scoring.

This work was supported by the Commission for Technology and Innovation (CTI) under grant No. 17988.1 PFIW-IW, the company Elite SA, and by the Swiss National Science Foundation (SNF) under grant No. 162809.

Author information

Authors and affiliations.

Sensory-Motor Systems Lab, Department of Health Sciences and Technology, Eidgenossische Technische Hochschule Zurich, Zurich, Switzerland

Elisabeth Wilhelm, Francesco Crivelli, Nicolas Gerig & Robert Riener

CSM SA, Center Alpnach, Robotics & Automation, Alpnach Dorf, Switzerland

Francesco Crivelli

Bio-Inspired RObots for MEDicine-Lab (BIROMED-Lab), Department of Biomedical Engineering, University of Basel, Basel, Switzerland

Nicolas Gerig

Department of Pulmonology, University Hospital Zurich, Zurich, Switzerland

Malcolm Kohler

Spinal Cord Injury Center, University Hospital Balgrist, Zurich, Switzerland

Robert Riener

You can also search for this author in PubMed   Google Scholar

Contributions

The study was designed by FC, RR, and MK. Device and sensor setup were developed by FC, EW, NG, and RR. Data processing and analysis was done by NG, MK and EW. The manuscript was written by EW. The manuscript was critically reviewed by all co-authors. The author(s) read and approved the final manuscript.

Corresponding author

Correspondence to Elisabeth Wilhelm .

Ethics declarations

Ethics approval and consent to participate.

The experimental procedure was approved by the Ethics Committee of ETH Zurich, no. EK 2015-N-70. Written informed consent was obtained from all subjects.

Consent for publication

The person depicted in figure one signed consent stating that the picture can be used in this publication.

Competing interests

This study was conducted within an industry collaboration project with Elite SA. Elite SA contributed to the design of the device and the study, but not to data collection and analysis, decision to publish, or preparation of the manuscript.

Additional information

Publisher’s note.

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . The Creative Commons Public Domain Dedication waiver ( http://creativecommons.org/publicdomain/zero/1.0/ ) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Cite this article.

Wilhelm, E., Crivelli, F., Gerig, N. et al. The anti-snoring bed - a pilot study. Sleep Science Practice 4 , 14 (2020). https://doi.org/10.1186/s41606-020-00050-2

Download citation

Received : 17 February 2020

Accepted : 13 August 2020

Published : 08 September 2020

DOI : https://doi.org/10.1186/s41606-020-00050-2

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Habitual snoring
• Intelligent bed
• Positional therapy

Sleep Science and Practice

ISSN: 2398-2683

• General enquiries: [email protected]

• ABOUT THE PROJECT
• ONGOING PROJECTS
• PUBLICATIONS
• SURVEY METHODS
• SPOTLIGHT ON:
• DRUGS AND ALCOHOL
• PHYSICAL ACTIVITY
• SEDENTARY BEHAVIOR
• SEE ALL FINDINGS
• WOMEN’S FINANCIAL SECURITY
• HOME OWNERSHIP
• EMERGENCY FUNDS
• HEALTH INSURANCE
• INVESTMENTS
• LIFE INSURANCE
• LONG TERM CARE
• VOLUNTEERISM
• CIVIC ENGAGEMENT
• RELIGIOUS ORGANIZATIONS
• WORKING FOR PAY
• AFFILIATE INTERVIEWS
• DIRECTORS’ BLOG POSTS
• EXPERT PERSPECTIVES
• PERSONAL PERSPECTIVES
• STANFORD CENTER ON LONGEVITY >
• Research Update on Diet
• Research Update on Sleep

Research Update 1

By Marie Conley Smith

I n a world full of opportunities, stressors, inequalities, and distractions, maintaining a healthy lifestyle can be challenging, and sleep is often the first habit to suffer. Good sleep hygiene is a huge commitment: it takes up about a third of the day, every day, and works best when kept on a consistent schedule. It does not help that the primary short-term symptoms of insufficient sleep can be self-medicated away with caffeine. However, the effects of sleep loss can range from inconvenient to downright dangerous; people have trouble learning and being productive, take risks more readily, and are more likely to get into accidents. These effects also last longer than it takes to get them, as recovering from each night of poor sleep takes multiple days. When it comes to sleep, every night counts. In this update, we will discuss what Stanford researchers have to say about sleep and why we need it, who is getting too little of it, and some of the latest findings that may help us sleep better.

We have not cracked the code on sleep

Despite this progress, scientists have not been able to crack the code of why sleep is critical to brain function. There is also little consensus about how sleep stages actually affect quality of sleep and how they affect us when we are awake.

Part of the challenge of cracking the code on sleep is how difficult it is to study. The gold standard of sleep study, polysomnography, developed by Dement in the 1960s, 1 is the most reliable tool for measuring many sleep characteristics and detecting sleep disorders such as obstructive sleep apnea and narcolepsy. However, it is expensive and time-consuming to run, which means that usually only a night or two is recorded. This snapshot of sleep may not reflect what normally occurs for a given person, and makes it difficult to draw conclusions about their behavior and performance in the days surrounding the sleep measurement.

The recent explosion in consumer wearable devices is a promising trend for researchers because of their potential to measure thousands of people’s sleep in their natural environments. They have not yet been widely adopted as measurement tools by scientists, however, as it is unclear if they provide the level of precision and measurement consistency required for a scientific study. Researchers at Stanford have called for these devices to be cleared by the FDA before using them to assign a diagnosis. 2 The “holy grail” would be a wearable device that could track sleep accurately while also providing performance information about the rest of the day, which would allow researchers to recognize more nuanced relationships between how people sleep and how it affects their lives.

The short- and long-term effects of insufficient sleep

We all know anecdotally what it is like to get too little sleep; it might be described with words and phrases like “tired,” “cranky,” “sluggish,” and “need caffeine.” Review of the scientific literature reveals how wide-ranging these effects can be. With too little sleep, people have a harder time learning 3 and concentrating, and are more likely to take risks. 4,5 The likelihood of getting into an auto accident increases. 6 Sleep deprivation has a bidirectional relationship with depression, 7,8 in that insomnia often both precedes and follows a depressive episode. Short sleep also interferes with other Healthy Living behaviors: people are more likely to crave sweet and fatty foods 9 and to choose foods that are calorically dense, 10 are more prone to injury during exercise, 11 and have an increased risk of obesity. 12

Sleep deprivation can even affect mundane daily activities. In 2017, then Stanford PhD candidate Tim Althoff and Professor Jamie Zeitzer of the Stanford Center for Sleep Sciences and Medicine took up the sleep measurement challenge by collaborating with Microsoft Research to examine the effects of sleep deprivation through a common daily activity: using an online search engine. 13 They paired users’ Microsoft Band sleep data with their Bing searches among users who had agreed to share their activity for study. By linking quantity and timing of sleep with typing speed during the searches, they were able to draw a number of conclusions about how sleep quality affects performance.

In this study, the researchers captured the sleep duration and search engine interactions of over 31,000 people. The researchers measured the amount of time between keystrokes as people typed their search engine entries, and used this as a measure of daily performance (that is, how well people did after a night of sleep). They were able to track the people who had multiple nights of insufficient sleep (defined as 6 hours of sleep or fewer) to see if their typing speed changed. They found that, on average, one night of insufficient sleep resulted in worse performance for three days, and two nights of insufficient sleep negatively impacted performance for six days. In other words, it took people almost an entire week to recover their performance after two consecutive nights of insufficient sleep. The implication is that the impact of sleep loss can persist for days.

Recent Stanford solutions for better sleep

Ongoing research at Stanford has led both to treatments for sleep disorders and to recommendations for best sleep practices for the public.

There are a few clinics and organizations that offer CBTI remotely in an effort to give more people access. There are apps such as SleepRate , which features content designed by Stanford researchers, Somryst , which was recently approved by the FDA, and Sleepio , which is offered by several large employers as an employee benefit. The Cleveland Sleep Clinic offers a 6-week online program called “ Go! to Sleep ,” and the U.S. Department of Veterans Affairs offers one of the same duration called “ Path to Better Sleep .” A physician should be consulted before starting any of these programs to ensure there are not any underlying disorders that need to be addressed.

Ultrashort light flash therapy Professor Jamie Zeitzer was interested in helping people who had a hard time sleeping because their circadian rhythm was not in sync with their desired sleep schedule. He discovered that ultrashort bursts of light directed into a person’s closed eyes while they were sleeping was very effective at shifting the time a person starts getting sleepy. Sleep doctors had already been using continuous light to help people reset their internal clock while they were awake; this new short-flash method shows great promise not only because of its effectiveness, but because it can be administered passively while people are sleeping. The approach involves wearing a sleep mask that emits the bright flashes and has been shown to only wake individuals who are particularly sensitive to light.

Lumos Sleep Mask

Professor Zeitzer and his team administered these ultrashort light flashes to teenagers, whose natural circadian systems have shifted so that their sleep and wake times are considerably later than children or adults. The time structure of our society, and schools in particular, does not take this into account. Professor Zeitzer administered the light flashes to see if it would help teens go to bed earlier. 20 They found that, while the teenagers were getting sleepy earlier, the light flashes alone were not enough to get the teenagers to bed earlier. With a second group of teens, they combined the light therapy with cognitive behavioral therapy (CBT) sessions. The CBT sessions served to inform the teens about sleep health and hygiene and helped them schedule their activities to allow for their desired sleep hours. After this combined therapy trial, the teens went to bed an average of 50 minutes earlier, getting an average of 43 more minutes of sleep per night. The researchers found the CBT component to be integral to behavior change – without the added education and support, the teens were not motivated enough to change their behavior and would simply push past their sleepiness.

This ultrashort light flash therapy can be used by anyone who may want to shift their sleep schedule; for example, to rebound from jet lag or to cope with a consistent graveyard shift at work. There is no evidence that other groups would require accompanying CBT like the teens, as long as they are self-motivated to change their sleep schedule. Zeitzer plans to test this technology next with older adults who wish to push their sleep time later. A company has spun out of this work, which Zeitzer advises but in which he has no financial interest, called Lumos . They are currently developing their product, and are hoping to make this intervention widely available.

Data Spotlight on: Black Americans

While most Americans have seen improvements in sleep over the past decade, Black Americans continue to sleep significantly less than other groups. This trend has been examined both by researchers and the popular press. 21,22 Researchers have found that Black Americans, in addition to getting shorter sleep, are also more likely to get poor quality sleep – spending less time in the most restorative stages of sleep 23,24 – and to develop obstructive sleep apnea. 25 Black Americans are also disproportionately affected by diseases that have been associated with poor sleep, such as obesity, diabetes, 26 and cardiovascular disease. 25

The exact reason(s) for Black Americans’ poor sleep is still unclear, though researchers have proposed potential contributing factors, largely related to the social inequality Black Americans face in the U.S.:

Experiences of discrimination : the stress of racial discrimination has been associated with spending lesstime in deep sleep and more time in light sleep among Black Americans. 24

Living environment : neighborhood quality has been linked to sleep quality, 27 and Stanford researchersfound that racial and income disparities persist in neighborhoods. 28 They found that while middle-income white families are more likely to live in resource-rich neighborhoods with other middle-income families, middle-income black families tend to live in markedly lower-income, resource-poorneighborhoods.

Work and income inequality : for example, shift work can cause irregular working hours. This leadspeople to suffer “social jetlag,”; a discrepancy in sleep hours between work and free days, 29 leading tosymptoms of sleep deprivation.

Lack of access to resources : particularly sleep-related healthcare and education.

Some of these factors are being addressed directly. Professor Girardin Jean-Louis from New York University and his team have devoted themselves to addressing the access to healthcare and education issue among local black communities in New York by tailoring online materials about obstructive sleep apnea to the culture, language, and barriers of specific communities. 30 Professor Jamie Zeitzer and his team at Stanford recently completed an initial clinical trial of a drug (suvorexant), which was found to help people who work at night get three more hours of sleep during the day. 31 Professor Zeitzer’s ultrashort light flash therapy (discussed above) may also help with shift work. These interventions could help to improve sleep for Black Americans, but they may not make up the whole picture; it could be that the underlying social inequality needs to be addressed in order to fully close the sleep gap.

Thanks to Jamie Zeitzer and Ken Smith for their insights and edits on this report.

• Deak, M., & Epstein, L. J. (2009). The history of polysomnography. Sleep Medicine Clinics , 4 (3), 313–321.
• Cheung, J., Zeitzer, J. M., Lu, H., & Mignot, E. (2018). Validation of minute-to-minute scoring for sleep and wake periods in a consumer wearable device compared to an actigraphy device. Sleep Science and Practice , 2 (1), 11. https://doi.org/10.1186/s41606-018-0029-8
• Gao, C., Terlizzese, T., & Scullin, M. K. (2019). Short sleep and late bedtimes are detrimental to educational learning and knowledge transfer: An investigation of individual differences in susceptibility. Chronobiology International , 36 (3), 307–318. https://doi.org/10.1080/07420528.2018.1539401
• O’Brien, E. M., & Mindell, J. A. (2005). Sleep and risk-taking behavior in adolescents. Behavioral Sleep Medicine , 3 (3), 113–133. https://doi.org/10.1207/s15402010bsm0303_1
• Rusnac, N., Spitzenstetter, F., & Tassi, P. (2019). Chronic sleep loss and risk-taking behavior: Does the origin of sleep loss matter? Behavioral Sleep Medicine , 17 (6), 729–739. https://doi.org/10.1080/15402002.2018.1483368
• Bioulac, S., Micoulaud-Franchi, J.-A., Arnaud, M., Sagaspe, P., Moore, N., Salvo, F., & Philip, P. (2017). Risk of motor vehicle accidents related to sleepiness at the wheel: A systematic review and meta-analysis. Sleep , 40 (10). https://doi.org/10.1093/sleep/zsx134
• Franzen, P. L., & Buysse, D. J. (2008). Sleep disturbances and depression: Risk relationships for subsequent depression and therapeutic implications. Dialogues in Clinical Neuroscience , 10 (4), 473–481.
• Tsuno, N., & Ritchie, K. (2005). Sleep and Depression. The Journal of Clinical Psychiatry , 16.
• Lv, W., Finlayson, G., & Dando, R. (2018). Sleep, food cravings and taste. Appetite , 125 , 210–216. https://doi.org/10.1016/j.appet.2018.02.013
• Pardi, D., Buman, M., Black, J., Lammers, G. J., & Zeitzer, J. M. (2017). Eating decisions based on alertness levels after a single night of sleep manipulation: A randomized clinical trial. Sleep , 40 (2). https://doi.org/10.1093/sleep/zsw039
• Chennaoui, M., Arnal, P. J., Sauvet, F., & Léger, D. (2015). Sleep and exercise: A reciprocal issue? Sleep Medicine Reviews , 20 , 59–72. https://doi.org/10.1016/j.smrv.2014.06.008
• Cappuccio, F. P., Taggart, F. M., Kandala, N. B., Currie, A., Peile, E., Stranges, S., & Miller, M. A. (2008). Meta-analysis of short sleep duration and obesity in children and adults.  Sleep ,  31 (5), 619-626.
• Althoff, T., Horvitz, E., White, R. W., & Zeitzer, J. (2017, April). Harnessing the web for population-scale physiological sensing: A case study of sleep and performance. In  Proceedings of the 26th international conference on World Wide Web (pp. 113-122).
• Roth, T. (2007). Insomnia: Definition, prevalence, etiology, and consequences. Journal of Clinical Sleep Medicine , 3 (5 Suppl), S7–S10.
• Qaseem, A., Kansagara, D., Forciea, M. A., Cooke, M., & Denberg, T. D. (2016). Management of chronic insomnia disorder in adults: A clinical practice guideline from the American College of Physicians. Annals of Internal Medicine , 165 (2), 125–133. https://doi.org/10.7326/M15-2175
• Jacobs, G. D., Pace-Schott, E. F., Stickgold, R., & Otto, M. W. (2004). Cognitive behavior therapy and pharmacotherapy for insomnia: A randomized controlled trial and direct comparison. Archives of Internal Medicine , 164 (17), 1888–1896. https://doi.org/10.1001/archinte.164.17.1888
• Manber, R., Bei, B., Simpson, N., Asarnow, L., Rangel, E., Sit, A., & Lyell, D. (2019). Cognitive behavioral therapy for prenatal insomnia: A randomized controlled trial. Obstetrics & Gynecology , 133 (5), 911–919. https://doi.org/10.1097/AOG.0000000000003216
• Ong, J. C., Crawford, M. R., Dawson, S. C., Fogg, L. F., Turner, A. D., Wyatt, J. K., Crisostomo, M. I., Chhangani, B. S., Kushida, C. A., Edinger, J. D., Abbott, S. M., Malkani, R. G., Attarian, H. P., & Zee, P. C. (2020). A randomized controlled trial of CBT-I and PAP for obstructive sleep apnea and comorbid insomnia: Main outcomes from the MATRICS study. Sleep . https://doi.org/10.1093/sleep/zsaa041
• Karlin, B. E., Trockel, M., Taylor, C. B., Gimeno, J., & Manber, R. (20130415). National dissemination of cognitive behavioral therapy for insomnia in veterans: Therapist- and patient-level outcomes. Journal of Consulting and Clinical Psychology , 81 (5), 912. https://doi.org/10.1037/a0032554
• Kaplan, K. A., Mashash, M., Williams, R., Batchelder, H., Starr-Glass, L., & Zeitzer, J. M. (2019). Effect of light flashes vs sham therapy during sleep with adjunct cognitive behavioral therapy on sleep quality among adolescents: A randomized clinical trial. JAMA Network Open , 2 (9), e1911944. https://doi.org/10.1001/jamanetworkopen.2019.11944
• Resnick, B. (2015, October 27). The Racial Inequality of Sleep . The Atlantic. https://www.theatlantic.com/health/archive/2015/10/the-sleep-gap-and-racial-inequality/412405/
• Resnick, B., & Barton, G. (2018, April 12). Black Americans don’t sleep as well as white Americans. That’s a problem. Vox. https://www.vox.com/science-and-health/2018/4/12/17224328/sleep-gap-black-white-minority-america-health-consequences
• Beatty, D. L., Hall, M. H., Kamarck, T. A., Buysse, D. J., Owens, J. F., Reis, S. E., Mezick, E. J., Strollo, P. J., & Matthews, K. A. (2011). Unfair treatment is associated with poor sleep in African American and Caucasian adults: Pittsburgh SleepSCORE project. Health Psychology , 30 (3), 351–359. https://doi.org/10.1037/a0022976
• Tomfohr, L., Pung, M. A., Edwards, K. M., & Dimsdale, J. E. (2012). Racial differences in sleep architecture: The role of ethnic discrimination. Biological Psychology , 89 (1), 34–38. https://doi.org/10.1016/j.biopsycho.2011.09.002
• Olafiranye, O., Akinboboye, O., Mitchell, J., Ogedegbe, G., & Jean-Louis, G. (2013). Obstructive sleep apnea and cardiovascular disease in blacks: A call to action from association of black cardiologists. American Heart Journal , 165 (4), 468–476. https://doi.org/10.1016/j.ahj.2012.12.018
• Jackson, C. L., Redline, S., Kawachi, I., & Hu, F. B. (2013). Association between sleep duration and diabetes in black and white adults. Diabetes Care , 36 (11), 3557–3565. https://doi.org/10.2337/dc13-0777
• Hale, L., Hill, T. D., & Burdette, A. M. (2010). Does sleep quality mediate the association between neighborhood disorder and self-rated physical health? Preventive Medicine , 51 (3–4), 275–278. https://doi.org/10.1016/j.ypmed.2010.06.017
• Reardon, S. F., Fox, L., & Townsend, J. (2015). Neighborhood income composition by household race and income, 1990–2009. The ANNALS of the American Academy of Political and Social Science , 660 (1), 78–97. https://doi.org/10.1177/0002716215576104
• Wittmann, M., Dinich, J., Merrow, M., & Roenneberg, T. (2006). Social jetlag: Misalignment of biological and social time. Chronobiology International , 23 (1–2), 497–509. https://doi.org/10.1080/07420520500545979
• Jean-Louis, G., Robbins, R., Williams, N. J., Allegrante, J. P., Rapoport, D. M., Cohall, A., & Ogedegbe, G. (2020). Tailored Approach to Sleep Health Education (TASHE): A randomized controlled trial of a web-based application. Journal of Clinical Sleep Medicine . https://doi.org/10.5664/jcsm.8510
• Zeitzer, J. M., Joyce, D. S., McBean, A., Quevedo, Y. L., Hernandez, B., & Holty, J.-E. (2020). Effect of suvorexant vs placebo on total daytime sleep hours in shift workers: A randomized clinical trial. JAMA Network Open , 3 (6), e206614. https://doi.org/10.1001/jamanetworkopen.2020.6614

Why your sleeping position is shortening your life

H ow many of us monitor our sleep posture? We have a favourite position; we fall asleep and that’s generally as far as it goes. But sleep position can have profound implications not only for the quality of sleep, but also for long-term health. Indeed, in the worst-case scenarios, a bad sleep posture may be slowly killing you.

Despite the impact sleep posture can have on conditions such as dementia and heart disease, research is limited and tends to focus on aches and pains. But back pain is just one implication of an unsuitable sleep position.

Chartered physiotherapist, sleep expert and author of The Good Sleep Guide , Sammy Margo, explains: “Sleep positions can significantly affect your overall health, comfort, and the quality of your sleep . Each position has its pros and cons and understanding them can help you make adjustments for better sleep and health outcomes.”

Dr Kat Lederle is a sleep scientist and the author of Sleep Sense . She points out that lifestyle factors in the day are usually the cause of postural problems at night.

“What you do in the day generally triggers the pain and discomfort that is felt when you sleep in certain positions. One of the most common contributing factors to this is a sedentary lifestyle, so it is important to move regularly during the day.”

The health risks of your sleep position – and how to mitigate them

Side sleeping.

Side sleeping is the most common position but there are health implications for certain people depending on whether they lie on their left or right side.

It is advised that pregnant women and anyone who suffers from acid reflux or gastroesophageal reflux disease (GERD) or other gut problems sleep on their left side.

“This is because the stomach is lower than your oesophagus,” explains Margo.

People with heart conditions , on the other hand, are advised to try sleeping on their right side to alleviate pressure on the heart. Studies show that when people lie on their left side the position of their heart shifts due to gravity. This causes changes in the heart’s electrical activity. Tissues and structures between the lungs hold the heart in place when you sleep on your right side.

Sleep position may also have an impact on brain health . During sleep the brain’s glymphatic system “washes” waste toxins away from the brain. There is evidence that suggests this process works better when we sleep on our right side.

“That is potentially of interest to people at risk of dementia or Alzheimer’s or any kind of neurodegenerative disease,” says Lederle.

Postural problems can occur with side sleepers depending on body shape, as Margo explains. “Women with hourglass figures sleeping on a soft mattress will sink into a banana shape and that will cause a strain on the spine and hips. While men who side-sleep can tend to get more pain in their shoulders as they get older and their muscles weaken.”

Side sleeping can also cause wrinkles and breast sagging because the skin on the face can get pressed against bedding and gravity can pull breast tissue and stretch skin.

One 2022 study by Beijing Forestry University and Chenzhou Vocational Technical College looked at the relationship between sleeping position and sleep quality. It used flexible wearable sensors to monitor sleep position and turning frequency. It concluded that subjects without sleep disorders who prefer to sleep on their side will sleep better than those who like to sleep on their back and that a higher frequency of turning during sleep will reduce sleep quality. 

Another study published in 2021 looked at relationships between sleep posture, back pain and quality of sleep. It reported that positions in which the spine was twisted can cause tissue microdamage and muscle spasms. The study compared common positions such as supine (back sleeping), provocative side lying (where the sleeper twists at the hip with one leg over the other), protected side lying (where the sleeper places a hand between the thighs and crosses the other arm over the chest), and prone (front sleeping).

It concluded that while it is not known if sleep posture is a risk factor for acute onset or recurrent back pain, participants with symptoms and stiffness in the morning spent more of the night in provocative (i.e. twisted at the hip) sleep postures.

To mitigate some of the problems associated with side sleeping Margo recommends using a thick pillow to align the head and neck with your spine and placing a pillow between your knees to support your hips and reduce strain on your lower back.

Back sleeping

One of the most common health problems associated with back sleeping is sleep apnoea, a condition whereby the soft tissue at the back of the throat relaxes and collapses the airway causing snoring and interrupted breathing.

Lederle explains: “This has implications for wider health and often goes hand in hand with obesity. It disrupts the continuity and quality of sleep. It can lead to tiredness, which can be a problem for people driving. There are also physical health implications. We know that poor quality sleep raises the risk of diabetes, heart disease and other comorbidities. Sleep apnoea opens the door to all these other conditions.”

One way to try to lessen the problem is to sleep in an elevated position.

However, for those who suffer from back and neck pain, back sleeping is often the best option.

Margo says: “The optimal position for spine alignment is lying on your back with a pillow under the knees to soften the back. This position preserves the natural contours of your spine. It can also minimise wrinkles.”

She also points out that for those who do not suffer from sleep apnoea, back sleeping can be a good position to train yourself into as you get older as back sleepers tend to have less back pain and back sleeping is also required for post-operative patients.

Front sleeping

While stomach sleeping may reduce snoring because it can help keep the airways more open than back sleeping, it is the position most likely to lead to increased neck and back pain.

“Twisting your neck to the side puts strain on your neck, and stomach sleeping can also arch your spine,” explains Margo.

“Direct pressure on the face can contribute to wrinkles over time,” she adds.

To help alleviate postural pain front sleepers are advised to use a thin pillow or no pillow at all to keep the neck in a more neutral position and to place a pillow under the pelvis to help keep the lower back supported.

How to change your sleep position

It is normal to move around at night, some people are more active than others and if you move, it is not always indicative of problematic sleep.

If you want to change your regular sleep position, gradually train yourself. For example, if you want to change from a back sleeper to a side sleeper lie on your favoured side for five minutes the first night and then roll onto your back. The following night increase to six minutes, then seven and so on. Start slowly and build up until you get used to the position.

Recommended

Why your pillow is ruining your sleep – and how to fix it

Sign up to the Front Page newsletter for free: Your essential guide to the day's agenda from The Telegraph - direct to your inbox seven days a week.

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

• View all journals
• My Account Login
• Explore content
• About the journal
• Publish with us
• Sign up for alerts
• Open access
• Published: 01 May 2024

Sleep pressure modulates single-neuron synapse number in zebrafish

• Anya Suppermpool   ORCID: orcid.org/0000-0002-2769-4896 1   nAff2 ,
• Declan G. Lyons   ORCID: orcid.org/0000-0003-1775-4653 1 ,
• Elizabeth Broom   ORCID: orcid.org/0009-0006-1439-4591 1 &
• Jason Rihel   ORCID: orcid.org/0000-0003-4067-2066 1  

Nature ( 2024 ) Cite this article

8601 Accesses

1 Citations

276 Altmetric

Metrics details

• Synaptic plasticity

Sleep is a nearly universal behaviour with unclear functions 1 . The synaptic homeostasis hypothesis proposes that sleep is required to renormalize the increases in synaptic number and strength that occur during wakefulness 2 . Some studies examining either large neuronal populations 3 or small patches of dendrites 4 have found evidence consistent with the synaptic homeostasis hypothesis, but whether sleep merely functions as a permissive state or actively promotes synaptic downregulation at the scale of whole neurons is unclear. Here, by repeatedly imaging all excitatory synapses on single neurons across sleep–wake states of zebrafish larvae, we show that synapses are gained during periods of wake (either spontaneous or forced) and lost during sleep in a neuron-subtype-dependent manner. However, synapse loss is greatest during sleep associated with high sleep pressure after prolonged wakefulness, and lowest in the latter half of an undisrupted night. Conversely, sleep induced pharmacologically during periods of low sleep pressure is insufficient to trigger synapse loss unless adenosine levels are boosted while noradrenergic tone is inhibited. We conclude that sleep-dependent synapse loss is regulated by sleep pressure at the level of the single neuron and that not all sleep periods are equally capable of fulfilling the functions of synaptic homeostasis.

Similar content being viewed by others

Whole-cortex in situ sequencing reveals input-dependent area identity

A brainstem–hypothalamus neuronal circuit reduces feeding upon heat exposure

Unsupervised restoration of a complex learned behavior after large-scale neuronal perturbation

Although sleep is conserved across the animal kingdom 1 , the precise functions of sleep remain unclear. As sleep deprivation leads to acute impairment of cognitive performance 5 , many theories posit that synaptic plasticity associated with learning and memory preferentially occurs during sleep 6 . For example, the synaptic homeostasis hypothesis (SHY) proposes that synaptic potentiation during wakefulness results in an ultimately unsustainable increase in synaptic strength and number that must be renormalized during sleep through synaptic weakening and pruning 2 , 7 , 8 . Such sleep-dependent renormalization has been postulated to broadly affect most excitatory synapses throughout the brain 2 .

Many, but not all, experimental observations of brain-wide changes in synapses have been consistent with the SHY. Globally, synaptic genes, proteins and post-translational modifications are upregulated during waking and renormalized during sleep 9 , 10 , 11 , 12 . In both flies and mice, the number and size of excitatory synapses also increase after prolonged waking and decline during sleep 3 , 10 , 13 . Long-term imaging of small segments of dendrites in young and adult mice has also been used to observe sleep–wake-linked synapse dynamics 4 , 14 , 15 and, in zebrafish, axon terminals of wake-promoting hypocretin neurons are regulated by the circadian clock 16 . However, other studies have observed no impact of sleep–wake states on synaptic strength and neuronal firing rates 17 , 18 , and some have observed synaptic strengthening during sleep 19 , 20 , 21 , 22 . Furthermore, distinct classes of synapse within the same neuronal population can be differentially regulated by sleep–wake states 23 , consistent with observations that synaptic plasticity can be regulated in a dendritic-branch-specific manner 24 . Together, these observations paint a complex picture of how sleep sculpts synapse number and strength, raising fundamental questions about whether sleep-dependent synaptic homeostasis operates uniformly across neuronal types and at which scale (for example, dendrite, neuron, circuit or population) sleep acts to modulate synapses.

To examine the scope and selectivity of sleep-linked synaptic plasticity, it is vital to comprehensively track the synaptic changes of individual neurons through sleep–wake states. To that end, we used in vivo synaptic labelling tools in larval zebrafish to image the same neurons and their synapses repeatedly over long timescales, enabling us to map single-neuron synapse changes across sleep and wake states.

Synapse counts change across 24 h

To visualize excitatory synapses in single zebrafish neurons, we adapted an established fibronectin intrabodies generated with mRNA display (FingR)-based transgenic system that selectively binds to and labels postsynaptic density protein 95 (PSD95) 25 , 26 , 27 , a major postsynaptic scaffold of excitatory synapses 28 , 29 and a readout of synaptic strength 30 , 31 , to enable simultaneous imaging of synapses and neuronal morphology (Fig. 1a ). Consistent with previous reports 25 , 27 , 32 , we confirmed that this modified FingR(PSD95) system labels synapses with high fidelity by driving expression of Tg(UAS:FingR(PSD95)-GFP-P2A-mKate2f) in the spinal cord with a Tg(mnx1:Gal4) driver line and co-labelling with anti-MAGUK antibodies that recognize the PSD95 protein family. Greater than 90% of FingR(PSD95) + puncta associated with MAGUK, while 100% of neuronal MAGUK puncta were co-labelled with FingR(PSD95) (Extended Data Fig. 1a–e,h–i ). The signal intensities of co-labelled MAGUK and FingR(PSD95) synapses were positively correlated, indicating that the signal intensity is a reliable readout of synaptic PSD95 content, as reported previously 26 (Extended Data Fig. 1f,g ).

a , The synapse labelling construct. Zinc finger (ZF) and KRAB(A) domains limit overexpression 25 . b , The strategy to sparsely label synapses of FoxP2.A + tectal neurons ( Methods ). c , Example FoxP2.A:FingR(PSD95) + neuron at 7 d.p.f., with the synapses (white arrowheads, left), nucleus (blue arrowheads, left) and membrane (magenta, right) co-labelled. d , Overnight time-lapse tracking of select synapses from the neuron in c . The normalized GFP intensity (shading) is shown for each synapse (rows). The complete neuron map is shown in Extended Data Fig. 2a . e , Larvae were raised on 14 h–10 h light–dark (LD) cycles (blue), constant light (LL, pink) or switched from LD to LL at 6 d.p.f. (free running (FR), green), and then imaged (arrows) ( Methods ). f , The average locomotor activity and 95% confidence intervals (CIs) of larvae reared under LD (blue, n  = 75), clock-break LL (pink, n  = 84) or FR (green, n  = 98) conditions. g – j , The mean and 68% CI (column 1) and individual neuron (columns 2–4) synapse counts ( g ), percentage change in synapse number calculated within each neuron ( h ), normalized synapse intensity ( i ) and percentage change in synapse intensity ( j ) under the LD (blue), LL (pink) or FR (green) conditions. For columns 2–4, a line is shown for each neuron, collected across 8 LD, 4 LL and 4 FR independent experiments. For h , synapse number change (Δ synapse number) dynamics are different during the day from those during the night under LD conditions (* P  = 0.043, repeated-measures analysis of variance (ANOVA)). Synapse number change dynamics under LD cycling are significantly different from those under LL conditions (* P  = 0.015, main effect of condition, two-tailed mixed ANOVA, post hoc Benjamini–Hochberg correction; Hedge’s g  = 0.761). For j , day–night dynamics are significantly different under LD from those under the other conditions ( P  < 0.01, repeated-measures ANOVA). Both daytime FR and LD day–night dynamics are significantly different from those under the LL condition (mixed ANOVA interaction (condition × time), P  = 0.029; FR versus LL, P  = 0.038, g  = 0.937; LD versus LL, P  = 0.027, g  = 0.792; post hoc Benjamini–Hochberg correction, two-tailed). At night, LD versus FR, g  = −0.538; LD versus LL, g  = −0.527. The diagram in a is adapted from ref.  27 , CC BY 4.0 , and the diagram in b is adapted from ref.  33 , CC BY 4.0 . The colour key in e applies also to f – i .

Source Data

To test whether behavioural state modulates synapse strength and number at the single-neuron level, we focused on larval tectal neurons, which are accessible to imaging, have well-defined morphological and functional identities 33 and have a stable window of synapse maturation from 7 to 9 days post-fertilization (d.p.f.) 34 . Tectal neurons also undergo spike-timing-dependent plasticity 35 and receive a mixture of inputs that foster ‘competition’ among synapses 36 , 37 , a criterion envisaged by the SHY 2 . To sparsely label tectal neurons, we co-electroporated a plasmid driving Gal4 off the foxp2.A promoter with tol2 mRNA into Tg(UAS:FingR(PSD95)-GFP-P2A-mKate2f) larvae at 3 d.p.f. 38 (Fig. 1b,c and Methods ). This method resulted in approximately 10% of larvae containing a single FoxP2.A:FingR(PSD95) + neuron, allowing for repeated, long-term imaging of the synapse counts and intensities in the same neuron in a continuously mounted preparation (Fig. 1c,d and Extended Data Fig. 2a ). After confirming the relative stability of tectal neuron synapse counts in the 6–9 d.p.f. developmental window (Extended Data Fig. 2b–d ), we imaged each labelled neuron across a 14 h–10 h light–dark cycle at 7 d.p.f., collecting images just after lights on (zeitgeber time 0 (ZT0), 7 d.p.f.), near the end of the day (ZT10) and after a night of sleep (ZT0, 8 d.p.f.) (Fig. 1e ; an example neuron with synapse changes tracked across two timepoints is shown in Extended Data Fig. 3 ), leaving larvae to behave freely between imaging sessions. On average, the tectal neuron synapse number increased significantly during the day from 137 to 153 synapses (+14.4%) but decreased at night by −1.90% to 146 synapses (Fig. 1g,h (blue)). Similar day–night changes in the net synapse counts were observed in separate experiments that imaged neurons over multiple days and nights (Extended Data Fig. 4a–e ), with no evidence of artefacts from repeated imaging (Extended Data Fig. 4f–h ). Moreover, the average synapse FingR(PSD95)–GFP signal intensity increased significantly during the waking day phase (+36.8%) and decreased in the night sleep phase (−11.7%) (Fig. 1i,j ).

To test whether these synaptic dynamics are influenced by the direct action of lighting conditions or are instead controlled by an internal circadian clock, we also tracked neurons under conditions of either constant light from fertilization, which prevents the formation of functional circadian clocks and leads to arrhythmic behaviour in zebrafish (clock-break) 39 , 40 , 41 , or constant light after light–dark entrainment, which maintains damped circadian behaviour (free running) 42 (Fig. 1e,f ). Under clock-break conditions, changes in synapse number and intensity were abolished and remained smaller compared with in larvae raised on light–dark cycles (Fig. 1g–j (pink)). Under free-running conditions, synapse numbers continued to increase during the subjective day and decrease during the subjective night, albeit strongly damped (Fig. 1g,h (green)). The average synapse intensity was significantly elevated across all timepoints and showed a further significant increase in strength only during the subjective day, with no loss of intensity during the subjective night (Fig. 1i,j (green)). Collectively, these data show that, while light influences the baseline levels of synaptic strength (Fig. 1i ), changes in synapse counts are independent of lighting conditions but do require an intact circadian clock (to drive rhythmic sleep–wake behaviour; see below) (Fig. 1g ).

Moreover, although rhythmic day–night changes in synapses were detected in the average of all of the single neurons, the tracking of individual neurons revealed that many cells have different, even opposing, synaptic dynamics (Fig. 1g–j (right)). We therefore sought to test whether these diverse patterns mapped onto distinct neuronal subtypes (that is, cellular diversity) or whether they are due to variations in animal behaviour (that is, individual sleep–wake histories).

Synapse cycling across neuronal subtypes

To test whether distinct synapse day–night dynamics are associated with morphological subtypes of tectal neurons, we measured position, branching, length and other parameters of FoxP2.A:FingR(PSD95)–GFP + neurons, many of which project only within the tectum at 7 d.p.f. Clustering analysis found four subtypes, consistent with previous studies 33 , 43 (Fig. 2a–c and Extended Data Fig. 5a–c ). Tracking synapses across three light–dark cycles revealed that each neuronal subtype has, on average, different patterns of net synapse counts (excluding the rarely observed type 1 neurons). Specifically, dynamics consistent with the SHY were robustly observed only in the densely bistratified type 2 neurons, with an average increase of 15.3 synapses during the day and a reduction of 17.7 synapses at night, and weakly observed in type 4 neurons (+8.5 during the day and −8.2 overnight; Fig. 2d–g and Extended Data Fig. 5d–f ). By contrast, many type 3 neurons consistently exhibited the opposite pattern, with an average increase in synapse number at night and a slight decrease during the day (Fig. 2d–g ). However, compared with under clock-break conditions, in which no subjective day–night-linked changes occur (Extended Data Figs. 5g–j and 6a,b ), the FingR(PSD95)–GFP signal intensity of type 3 and 4 neurons, but not type 2 neurons, increased during the day and decreased at night (Extended Data Fig. 6a–c ), suggesting that synapse number and PSD95 content are differentially regulated in tectal subtypes. These subtype-specific alterations in synapse number cannot be explained by differences in larval sleep–wake behaviour, as the sleep amount was the same regardless of which neuron subtype was labelled in the larva (Extended Data Fig. 7a–c ).

a , The morphological parameters used to characterize FoxP2.A tectal neurons. A–P, anterior–posterior. b , Examples of each morphological subtype, chosen from n  = 17 (type 1), n  = 28 (type 2), n  = 61 (type 3) and n  = 42 (type 4) neurons collected over 26 independent experiments. The blue circles label nuclei. c , Example of the parameters used to distinguish the four subtypes. For the box plots, the centre lines show the median, the box limits show the interquartile range and the whiskers represent the distribution for each parameter. The slashed zero indicates that the feature is absent. See also Extended Data Fig. 5 . d – g , Synapse counts across multiple LD cycles for FoxP2.A tectal neurons of different subtypes. d , e , Average (68% CI) synapse counts ( d ) and average (68% CI) synapse number change ( e ) of subtypes (column 1) and for each neuron (columns 2–4), collected over 8 independent experiments. f , g , Average (68% CI) synapse counts ( f ) and net change ( g ), averaged across all days and nights for each subtype and larvae, including additional neurons tracked over a single day (Extended Data Fig. 5 ). Tectal subtype influences synapse changes (mixed ANOVA, interaction P  = 0.012, subtype × time). Type 2 ( n  = 16) and type 4 ( n  = 15) neurons gain more synapses during the day under LD conditions compared with under LL clock-break conditions ( P  = 0.018, g  = 0.952; P  = 0.021, g  = 0.812, respectively). At night, both type 2 and type 4 neurons lose synapses relative to type 3 (type 2 versus type 3, P  = 0.038; g  = −0.714; type 4 versus type 3, P  = 0.038, g  = −0.781, post hoc Benjamini–Hochberg correction, one-tailed). For b , scale bars, 10 μm.

As type 2 neurons have two prominent arbourization fields, we examined whether changes in day–night synapse number are heterogenous across different dendritic segments of individual neurons. Analysing the synapse number changes in four distinct classes of dendritic segment in type 2 neurons revealed that only the proximal arbour, which receives local inputs from the tectum and long-range inputs from brain areas such as the hypothalamus 44 , displayed significantly robust average increases in synapse number during the day and reductions at night (Extended Data Fig. 7d–f ). By contrast, synapse number dynamics within the distal arbour, which receives the majority of its inputs from the retina 43 , were more diverse. No correlations could be detected among the different dendritic compartments within the same neuron (Extended Data Fig. 7f ), suggesting that the time of day and sleep–wake states do not have uniform effects on synapse number even within the same neuron.

Sleep pressure facilitates synapse loss

If the synapses of individual neurons are regulated by sleep–wake states independently of the circadian clock, these dynamics should be altered by sleep deprivation (SD). We developed a gentle handling SD protocol in which zebrafish larvae are manually kept awake with a paintbrush for 4 h at the beginning of the night (ZT14–ZT18) and subsequently allowed to sleep (Supplementary Video  1 ). Sleep in larval zebrafish is defined as a period of inactivity lasting longer than 1 min, as this is associated with an increased arousal threshold, homeostatic rebound and other criteria of sleep 40 , 45 . After SD, the phase of the circadian clock machinery was unaffected, but larvae slept significantly more, with individual sleep bouts lasting longer, compared with non-sleep-deprived larvae (Extended Data Fig. 8a,b ), consistent with SD leading to increased sleep pressure 46 , 47 , 48 . Next, we visualized synapses of individual tectal neurons at 7 d.p.f. immediately before (ZT13–ZT14) and after (ZT18–ZT20) SD, and again the next morning (ZT0–ZT1) (Fig. 3a and Extended Data Fig. 9a ). Between the imaging sessions, we used video tracking to monitor sleep–wake behaviour ( Methods ). In control larvae, tectal neurons lost synapses overnight; however, this synapse loss was confined to the first part of the night (ZT14–ZT18), with an average loss of 1.7 synapses per hour, in contrast to the last part of the night (ZT18–ZT24), during which synapse loss was undetectable (+0.2 synapses per hour) (Fig. 3b (blue)). By contrast, neurons gained an average of 2.8 synapses per hour during SD (Fig. 3b (orange)). During the recovery period after SD, tectal neurons lost synapses at a rate of 2.2 synapses per hour (Fig. 3b and Extended Data Fig. 8c ). As during normal sleep, FoxP2.A tectal neuron subtypes responded differently to SD, with type 2 and even type 3 neurons (which did not have SHY-concordant changes under baseline conditions) gaining synapses during SD and losing them during recovery sleep, whereas type 4 neurons did not show any change (Extended Data Fig. 8d ). This suggests that SD biases synapses towards loss during subsequent sleep, even in neurons with different synapse dynamics under baseline conditions.

a , The 4 h gentle handling SD paradigm (ZT14–ZT18). Larvae were video-tracked and neurons were periodically imaged (arrows). b , The mean ± s.e.m. change in synapse counts per hour for the SD (orange, n  = 31 neurons) and control (blue, n  = 28) groups. c , Sleep time versus the change in synapse counts per hour for each larva during either the early (ZT14–ZT18, left) or late (ZT18–ZT24, middle) night for controls and after SD (ZT18–ZT24, right). The rate of synapse change is negatively correlated with sleep time during both early and late night but not after SD. d , In control larvae, the change in early night synapse counts is negatively correlated with late night synapse change. Early and late sleepers are defined as larvae that either sleep more in the first or second phase of the night, respectively. e , Synapse counts per hour for early- and late-night sleeping control larvae in the early (ZT14–ZT18) and late (ZT18–ZT24) phases of the night. Data are mean ± s.e.m. f – h , The reticulospinal neuron synapse number is modulated by sleep and wake states. f , Example reticulospinal neurons from the Tg(pvalb6:KALTA4) u508 line co-labelled by FingR(PSD95)–GFP (green, nuclei and synapses) and mKate2f (magenta, membrane). Vestibulospinal (VS) and MiD2cm neurons are indicated by the dashed ovals. g , Vestibulospinal (top) and MiD2cm (bottom) neurons from different larvae showing FingR(PSD95) + synapses (green) co-localized to the cell membrane (magenta). h , Changes in synapse number (mean and 68% CI) from ZT14 to ZT18 for vestibulospinal and MiD2cm neurons. Each dot represents the average across multiple neurons per larva. For b and e , statistical analysis was performed using two-tailed mixed ANOVA interaction (condition × time) with post hoc Benjamini–Hochberg correction; ****P  = 0.00007, *** P  = 0.0002 and ** P  = 0.006 ( b ) and * P  = 0.01 ( e ). For h , statistical analysis was performed using one-tailed Student’s t -tests; * P  < 0.03. Scale bars, 15 μm ( f ) and 10 μm ( g ). The lines in c and d depict the linear regression with the 95% CI.

As both SD and control larvae were at the same circadian phase, we conclude that sleep–wake states are the main driver of net changes in synapses in tectal neurons, and the effects of circadian clock disruption on synapses were primarily due to the loss of sleep rhythms (Fig. 1 ). Consistent with this interpretation, the total time that each larva spent asleep was significantly correlated with the rate of synapse change (Fig. 3c and Extended Data Fig. 8g ). Only after SD, when sleep and synapse loss were high across most larva–neuron pairs, was this correlation lost, which may indicate that either the machinery that supports sleep-dependent synapse loss can saturate or SD-induced rebound sleep is not fully equivalent to baseline sleep. The converse relationship was not observed, as the rate of synapse gain during SD was not correlated with either the subsequent total sleep or the average sleep bout lengths of single larvae (Extended Data Fig. 8f ). Consistent with the effects of SD, natural individual variation in sleep timing was predictive of the time period in which synapses were lost. ‘Early sleepers’ slept more during the first half of the night and lost synapses only during this period, whereas ‘late sleepers’ preferentially slept in the second half of the night and had a net loss of synapses only during the late night (Fig. 3d,e and Extended Data Fig. 8e ). Finally, to test whether sleep-dependent synapse loss is generalizable to neurons that do not receive direct retinal input, we confirmed that synapses of both presumptive vestibulospinal neurons that stabilize posture 49 and MiD2cm reticulospinal neurons involved in fast escapes 50 , 51 showed synapse gains during SD and synapse loss during sleep (Fig. 3f–h ).

Two explanations are consistent with the observed relationships between sleep and synapse change: either sleep is a permissive state for synapse loss, or sleep pressure, which builds as a function of waking, drives synapse loss during subsequent sleep. As sleep pressure and subsequent sleep amount at night are tightly linked under both baseline and SD conditions, we sought to disentangle their relative influences on synaptic change using sleep-inducing drugs to force larvae to sleep during the day, when sleep pressure remains low (Fig. 4a,b and Extended Data Fig. 9c ). Exposing larvae for 5 h during the day (ZT5–ZT10) to either 30 µM melatonin, which in zebrafish is a natural hypnotic that acts downstream of the circadian clock to promote sleep 52 , or 30 µM clonidine, an α2-adrenergic receptor agonist that inhibits noradrenaline release and increases sleep in zebrafish 45 , 53 , significantly and strongly increased total sleep and the average length of sleep bouts mid-day (Fig. 4c and Extended Data Fig. 10a,b ), with this drug-induced sleep remaining reversible by strong stimuli (Extended Data Figs. 9d,e and 10d–g ). Forced daytime sleep altered the build-up of sleep pressure, leading to reduced and delayed sleep in the subsequent night (Extended Data Fig. 9e ). However, drug-induced sleep at a time of low sleep pressure was not sufficient to trigger synapse loss, with tectal neurons still gaining an average of 1.0–1.7 synapses per hour, which was not significantly different from the synapse gains in the controls (Fig. 4d ). Similarly, artificially boosting adenosine signalling—one of the postulated molecular substrates of sleep pressure 54 —by administering 45 µM 2-choloroadenosine increased sleep during the day but also led to net gains in tectal neuron synapses (+0.9 synapse per hour) (Fig. 4c and Extended Data Fig. 10c ). Tectal neurons also gained synapses (+0.4 synapse per hour) in larvae that were co-administered 2-chloroadenosine and melatonin, despite sleeping more than 35 minutes per hour (Fig. 4c,d ). By contrast, simultaneously boosting adenosine signalling while inhibiting noradrenaline release with clonidine resulted in synapse loss (−0.8 synapses per hour) in tectal neurons (Fig. 4c,d and Extended Data Fig. 9c ), which express both adenosine and adrenergic receptors (Extended Data Fig. 11 ). These results demonstrate that daytime sleep can support synapse loss under conditions of high sleep pressure and low noradrenergic tone, possibly through direct signalling events.

a , Larvae were temporarily treated with sleep-promoting drugs during the day (ZT5–ZT10). The black arrows indicate the imaging periods before and after drug treatment. b , Drug-induced sleep during the day disentangles sleep pressure (that is, low) from sleep amount (that is, high), which are otherwise tightly correlated. c , Drug-treated larvae sleep significantly more during the day compared with the dimethyl sulfoxide (DMSO)-treated controls. d , During the day (from ZT5–ZT10), synapse counts increase under all control and drug conditions, except during co-administration of clonidine and 2-chloroadenosine, when synapses are significantly lost. Data are mean ± s.e.m. n values represent the number of neurons (top row) or fish (bottom row). For c and d , statistical analysis was performed using Kruskal–Wallis tests with post hoc Dunn’s multiple-comparison test (left) and one-way ANOVA (right); not significant (NS), P  > 0.5; * P  = 0.034, ** P  < 0.01, **** P  < 0.0001.

The SHY proposes that synapse numbers and strength increase during wake and decrease during sleep. By tracking synapses of single tectal neurons through sleep–wake states and circadian time, our data resolve several outstanding questions about the scale, universality and mechanisms of sleep-linked plasticity. We show that SHY-concordant dynamics of the synapse population within single neurons are present on average across many cells but, when examined on a neuron-by-neuron basis, more diverse patterns of synapse change are revealed. These observations may explain some discrepancies among previous studies of the SHY, as these single-neuron synaptic dynamics would not be captured by population-level, single-time-point snapshots of synapse number or function. We also show that sleep is necessary but not sufficient for synaptic loss, as synapse loss occurred only when sleep was accompanied by high sleep pressure associated with adenosine signalling and low noradrenergic tone. Adenosine signalling has been shown to promote Homer1a-dependent downscaling and destabilization of synapses, whereas noradrenergic signalling has been found to prevent this process 55 . Our data link these mechanisms to sleep pressure and sleep behaviour in vivo. Whether single-neuron or subcellular variation in the expression or sensitivity to these synapse-regulating signals could account for the diversity of synapse alterations remains an interesting possibility for future work. Sleep pressure, as reflected by the density of slow-wave activity in mammalian sleep, has also been linked to changes in synapses associated with learning and memory 11 , 56 . We find that sleep-linked synapse loss depends on molecular signals linked to high sleep pressure and, notably, also mirrors slow-wave activity by occurring predominantly in the early part of the sleep period 6 . This finding raises the question of whether epochs of sleep associated with low sleep pressure, such as in the latter half of the night, have additional, non-synaptic remodelling roles. If so, the evolution, persistence and ubiquity of these different sleep epochs could be under specific regulatory and selective pressures.

Zebrafish husbandry and experiments were conducted according to UCL Fish Facility standard protocols and under project licenses PA8D4D0E5 and PP6325955 awarded to J.R., according to the UK Animal Scientific Procedures Act (1986). Embryos were kept in Petri dishes in fish water (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl 2 , 0.33 mM MgSO 4 and 0.1% methylene blue) in a 14 h–10 h light–dark cycle incubator at 28 °C. Petri dishes exposed only to fish water were cleaned with 75% ethanol, washed, soaked overnight in distilled water, air-dried and rinsed with fish water before reuse. The sex of AB/TL zebrafish larvae is not biologically determined at the early developmental stages used for these studies.

Cloning and transgenesis

Transgene constructs that simultaneously encode FingR targeting PSD95 and membrane markers of neuronal morphology were generated using the In-Fusion HD Cloning System (Clontech). First, the GFP in a pCS2-P2A-GFP-CAAX was replaced with mKate2f by combining the linearized pCS2 (through inverse PCR; primers: 5′-GGATCTAGGACCGGGGTTTTC-3′ and 5′-GTGCTCTCCTGACCTCTAGAA-3′) with amplified mKate2f from dUAS-mKate2f (gift from the Tada laboratory, UCL) with 15 bp overhangs complementary to pCS2 site of insertion (primers: 5′-CCCGGTCCTAGATCCATGGTGAGCGAGCTGATTAAG-3′ and 5′- AGGTCAGGAGAGCACTCAGGAGAGCACACAGCAGCT-3′). Next, the template plasmid pTol2-zcUAS:PSD95.FingR-EGFP-CCR5TC-KRAB(A) (from the Bonkowsky laboratory, University of Utah; Addgene, 72638) was linearized by inverse PCR after the KRAB(A) sequence (primers: 5′-AGCCATAGAAGCAAGATTAGA-3′ and 5′ - GGAGGTGTGGGAGGTTTTTTC - 3′). The P2A-mKate2f sequences were then amplified with 15 bp overhangs complementary to the pTol2-zcUAS:PSD95.FingR-EGFP-CCR5TC-KRAB(A) insertion site (primers: 5′-CTTGCTTCTATGGCTGCCACGAACTTCTCTCTGTTA-3′ and 5′- ACCTCCCACACCTCCTCAGGAGAGCACACAGCAGCT-3′) and combined with the linearized FingR template.

To generate the stable Tg(UAS:FingR(PSD95 ) -GFP-CCR5TC-KRAB(A )- P2A-mKate2f ) line, purified pTol2-zcUAS:PSD95.FingR-EGFP-CCR5TC-KRAB(A)-P2A-mKate2f DNA construct was sequenced to confirm gene insertion and co-injected (10 ng µl −1 ) with emx3:Gal4FF 57 (10 ng µl −1 ) and tol2 transposase mRNA (100 ng µl −1 ) at 1 nl into wild-type TL embryos at the one-cell stage. At 3 d.p.f., injected embryos were screened for mosaic expression of mKate2f, then raised to adulthood. The tol2 transposase mRNA was in vitro transcribed from the NotI-linearized pCS-TP6287 plasmid (gift from the Wilson laboratory, UCL) using the SP6 mMESSAGE mMACHINE Kit (Ambion). RNA was purified using RNA Clean & Concentrator Kits (Zymo Research). Germline transmission was determined by mating adult fish to nacre mutants ( mitfa w2 / w2 , pigmentation mutants 58 ) and subsequently identifying their progeny for mKate2f fluorescence, then raising to adulthood to establish a stable Tg ( UAS:FingR ( PSD95 ) -GFP-CCR5TC-KRAB ( A ) -P2A-mKate2f ) u541 ; Tg ( emx3:Gal4FF ) u542 line. Owing to the negative-feedback mechanism in the system, Tg ( UAS:FingR ( PSD95 ) -GFP-CCR5TC-KRAB ( A ) -P2A-mKate2f ) expression is extremely low. To increase the number of transgene copies and the level of expression in the background reporter line, the double transgenic Tg ( UAS:FingR ( PSD95 ) -GFP-CCR5TC-KRAB ( A ) -P2A-mKate2f ) ;Tg ( emx3:Gal4 ) fish were incrossed for imaging experiments and maintained by alternating incrosses and outcrosses to nacre mutants.

Whole-mount synaptic immunohistochemistry and imaging

Staining for MAGUK expression was performed using whole-mount immunohistochemistry adapted from a previous study 59 . Zebrafish larvae (2 d.p.f.) were dechorionated and fixed with 4% formaldehyde methanol-free (Pierce Thermo Fisher Scientific, 28906) in BT buffer (1.0 g sucrose, 18.75 µl 0.2 M CaCl 2 , topped up to 15 ml with PO 4 buffer (8 parts 0.1 M NaH 2 PO 4 and 2 parts 0.1 M Na 2 HPO 4 )). To increase the signal-to-noise ratio, the fixing time was decreased to 1.5–2 h at 4 °C, although this led to softer samples. The samples were washed with PO 4 buffer and distilled H 2 O for 5 min at room temperature, then permeabilized with ice-cold 100% acetone for 5 min at −20 °C. After washing with distilled H 2 O and PO 4 buffer for 5 min each, the samples were blocked with blocking buffer containing 2% goat serum, 1% bovine serum albumin and 1% DMSO in 0.1 M PBS pH 7.4 for at least 2 h. The samples were then incubated with primary antibodies (see below for list) diluted in blocking buffer at 4 °C overnight. The embryos were washed 4–6 times for at least 20 min in blocking buffer at room temperature and incubated in secondary antibodies overnight at 4 °C. To remove unbound secondary antibodies, the embryos were washed again and transferred to glycerol in a stepwise manner up to 80% glycerol in PBS.

The primary antibodies used for staining were anti-pan-MAGUK (mouse monoclonal, K28/86, Millipore) and anti-tRFP (rabbit polyclonal, AB233, Evrogen), both at a dilution of 1:500. To avoid overamplification of signal outside of the synapse, FingR(PSD95)–GFP puncta were visualized using its own fluorescence. The following secondary antibodies were used at a dilution of 1:200: Alexa-Fluor 568 goat anti-rabbit IgG and Alexa-Fluor 633 goat anti-mouse IgG monoclonal (Life Technologies).

Confocal images were obtained using the Leica TCS SP8 system with HC PL APO ×20/0.75 IMM CS2 multi-immersion objective set to glycerol (Leica Systems). z stacks were obtained at 1.0 μm depth intervals with sequential acquisition settings of 1,024 × 1,024 px. The raw images were compiled using NIH Image J ( http://imagej.nih.gov/ij/ ). To analyse the colocalization of the puncta, maximum projections of 5–10 μm were taken for each cell. Grey values were taken from the cross-section of the puncta using the plot-profile tool from ImageJ. Puncta grey values were normalized against the whole-stack grey value of their respective channels.

The colocalization and relationships between FingR(PSD95)–GFP and antibody staining were analysed using custom Python scripts (available at GitHub ( https://github.com/anyasupp/single-neuron-synapse )). For colocalization of FingR and antibody puncta (and vice versa), the presence of puncta with maximum normalized grey value of at least 50% higher than the baseline was used. To estimate the size of the puncta, the normalized grey values were interpolated with a cubic polynomial implemented by the SciPy (v.1.11.4) function scipy.interpolate.interp1d before finding the full width at half maximum.

Single-cell FingR(PSD95) expression using electroporation

To sparsely label single tectal cells, a FoxP2.A:Gal4FF activator plasmid (gift from M. Meyer) was electroporated into the Tg ( UAS:FingR ( PSD95 ) -GFP-ZFC ( CCR5TC ) -KRAB ( A ) -P2A-mKate2f )-positive larvae at 3 d.p.f according to a previously described method 33 . Anaesthetized 3 d.p.f. zebrafish larvae were mounted in 1% low-melting-point agarose (Sigma-Aldrich), perpendicular to a glass slide in a Petri dish filled with electroporation buffer (180 mM NaCl, 5 mM KCl, 1.8 mM CaCl 2 , 5 mM HEPES, pH 7.2) with 0.02% tricaine (MS-222, Sigma-Aldrich). Excess agarose along the larval body was then removed to allow access for the electroporation electrodes. A FoxP2.A:Gal4FF construct (500 ng µl −1 ) was injected into the midbrain ventricle together with tol2 mRNA (20 ng µl −1 ) and Phenol Red (~0.025%) at 5–8 nl using a micro glass needle (0.58 mm inside diameter, Sutter Instrument, BF100-58-15) pulled using a micropipette puller (Model P-87 Sutter Instrument). After injection, the positive electroporation electrode was placed lateral and slightly dorsal to the hemisphere of the target optic tectum, and the negative electrode was placed lateral and ventral to the contralateral eye. Five 5 ms trains of 85 V voltage pulses at 200 Hz were delivered through the electrodes using an SD9 stimulator (Grass Instruments). Electroporated larvae were screened for sparse, single-cell expression of FoxP2:FingR(PSD95) + neurons using a ×20/1.0 NA water-dipping objective and an LSM 980 confocal microscope with Airyscan 2 (Zeiss) at 5–6 d.p.f.

Repeated Imaging of FingR-labelled synapses

For synapse-tracking experiments, Tg ( UAS:FingR ( PSD95 ) -GFP-CCR5TC-KRAB ( A ) -P2A-mKate2f ) larvae that were electroporated with FoxP2.A:Gal4FF were reared at 28 °C under various light schedules. At 5–6 d.p.f., larvae were visually screened for the expression of single or sparsely labelled FoxP2.A:FingR(PSD95) + neurons in the tectum using a ×20/1.0 NA water-dipping objective and the LSM 980 confocal microscope with Airyscan 2 (Zeiss) and placed into individual wells of six-well plates (Thermo Fisher Scientific) to keep track of individual larvae and the corresponding labelled neurons, each well containing approximately 10 ml of fish water. For repeated live imaging of reticulospinal neurons, Tg ( UAS:FingR ( PSD95 ) -GFP-CCR5TC-KRAB ( A ) -P2A-mKate2f ) were crossed to a Tg ( pvalb6:KALTA4 ) u508 driver line 50 (gift from the Bianco laboratory at UCL) and visually screened for larvae with a labelled reticulospinal population. For imaging FingR(PSD95)-GFP puncta, the larvae were anaesthetized with 0.02% tricaine for 5–10 min and immobilized in 1.5–2% low-melting-point agarose (Sigma-Aldrich) in fish water. The larvae were head-immobilized with the tail free and allowed to recover from anaesthesia during imaging. Imaging was performed at the appropriate zeitgeber/circadian time (ZT, where ZT0 is lights on) according to the experimental paradigm. For day–night synapse tracking, larvae were repeatedly imaged at approximately ZT0–ZT2 and ZT10–ZT12 at 7 d.p.f., 8 d.p.f. and 9 d.p.f. at 28.5 °C with the chamber lights on. For imaging performed during the dark phase (ZT14–ZT24), the temperature was kept at 28.5 °C with the chamber lights off. When immobilizing the larvae for night imaging, the handling was performed under low red light (Blackburn Local Bike Rear Light 15 Lumen; 5.2–30.5 lux, measured at the plate level). After imaging, larvae were unmounted from agarose by releasing agarose around their heads and allowing the larvae to independently swim out of the agarose. Unmounted larvae were then placed back into individual wells of six-well plates.

FingR(PSD95) + neuron image stacks were acquired using a ×20/1.0 NA water-dipping objective and the LSM 980 confocal microscope with Airyscan 2 (Zeiss). GFP and mKate2f were excited at 488 nm and 594 nm, respectively. z stacks were obtained at a 0.34 μm voxel depth with sequential acquisition settings of 2,024 × 2,024 px, giving a physical resolution of 0.0595376 μm in x , 0.0595376 μm in y and 0.3399999 μm in z and 16-bit using SR4 mode (imaging 4 pixels simultaneously). Pixel alignment and processing of the raw Airyscan stack were performed using ZEN Blue software (Zeiss).

Locomotor activity assay

Tracking of larval zebrafish behaviour was performed as previously described 45 , with slight modifications. Zebrafish larvae were raised at 28.5 °C under a 14 h–10 h light–dark (LD) cycle or constant light (LL) or switching from 14 h–10 h light–dark to constant light (free-running (FR) conditions). At 5–6 d.p.f., each FoxP2.A:FingR(PSD95) + larva was placed into individual wells of a six-well plate (Thermo Fisher Scientific) containing approximately 10 ml of fish water. The locomotor activity of some larvae was monitored using an automated video tracking system (Zebrabox, Viewpoint LifeSciences) in a temperature-regulated room (26.5 °C) and illuminated with white lights on either 14 h–10 h light–dark cycles or constant light conditions at 480–550 lux with constant infrared illumination. The larval movement was recorded using the Videotrack ‘quantization’ mode with the following detection parameters: detection threshold, 15; burst, 100; freeze, 3; bin size, 60 s. The locomotor assay data were analysed using custom MATLAB (MathWorks) scripts available at GitHub ( https://github.com/JRihel/Sleep-Analysis ). Any 1 min period of inactivity was defined as 1 min of sleep, according to the established convention for larval zebrafish 40 . For experiments examining the effects of drug treatment on behaviour that did not involve live imaging, such as the clonidine dark pulse experiment (Extended Data Fig. 10d–g ), 24-well (Thermo Fisher Scientific) and 96-well plates (Whatman) were used instead of the 6-well plates used for synapse imaging experiments. Sleep latency for Extended Data Fig. 9c–e was calculated using frame-by-frame data (collected at 25 fps), using code available at GitHub ( https://github.com/francoiskroll/FramebyFrame ).

Sleep deprivation assay

Zebrafish larvae were raised at 28.5 °C under a 14 h–10 h light–dark cycle to 6 d.p.f., when they were video-tracked (see the ‘Locomotor activity assay’ section). Randomly selected 7 d.p.f. larvae were then sleep deprived for 4 h immediately after lights off from ZT14 to ZT18. Non-deprived control larvae were left undisturbed. Larvae that were individually housed in six-well plates were manually sleep deprived under dim red light (Blackburn Local Bike Rear Light 15 Lumen) by repeated gentle stimulation using a No. 1-2 paintbrush (Daler-Rowney Graduate Brush) to prevent larvae from being immobile for longer than 1 min. For most stimulations, this required only putting the paintbrush into the water; if the larvae remained immobile, they were gently touched. The 4 h SD protocol was performed by experimenters in 2 h shifts. All sleep deprived and control larvae were imaged at around ZT14 and ZT18 on 7 d.p.f. and again at ZT0 on 8 d.p.f. (see the ‘Repeated imaging of FingR-labelled synapses’ section).

Drug exposure for live imaging

Tg ( UAS:FingR ( PSD95 ) -GFP-CCR5TC-KRAB ( A ) -P2A-mKate2f ) larvae that had been electroporated with FoxP2.A:Gal4FF (see the ‘Single-cell FingR(PSD95) expression using electroporation’ section) were kept under a 14 h–10 h light–dark cycle until 7 d.p.f., then imaged at ZT4–ZT5 (see the ‘Repeated imaging of FingR-labelled synapses’ section). Larvae were transferred to individual wells of a six-well plate containing 10 ml of sleep-promoting drugs, alone or in combination, as follows: 30 µM melatonin (M5250, Sigma-Aldrich) in 0.02% DMSO; 30 µM of clonidine hydrochloride (C7897, Sigma-Aldrich) in 0.02% DMSO; 45 µM 2-chloroadenosine (C5134, Sigma-Aldrich) in 0.02% DMSO; and 0.02% DMSO in fish water as controls 45 , 52 , 60 , 61 . Combinations of drugs were applied at the same concentrations as the single-dose conditions, maintaining the final DMSO concentration of 0.02%. Sleep induction was monitored with video-tracking (see the ‘Locomotor activity assay’ section) for 5 h, after which the drugs were removed by 2–3 careful replacements of the fish water using a transfer pipette followed by transferring the larvae individually to a new six-well plate with fresh water. The larvae were then reimaged using the Airyscan system (see the ‘Repeated imaging of FingR-labelled synapses’ section).

Tectal cell segmentation and clustering

The morphology of tectal neurons at 7 d.p.f. was segmented and measured using Imaris v.8.0.2 (Bitplane) and ImageJ (NIH). The total filament length for each neuron was obtained using the Imaris Filament function. The anterior–posterior span of the distal arbour was calculated using the Measurement function at an orthogonal view in 3D. The relative proximal arbour locations were calculated by dividing the proximal arbour distance from the nucleus by the total length of the neuron obtained using Filament function of Imaris. The distance from the skin, distal arbour thickness and distal arbour to skin distance were obtained using the rectangle Plot_Profile tool of ImageJ at an orthogonal view of the neuron to calculate the fluorescence intensity across the tectal depth. The intensity profiles were then analysed using custom Python scripts to obtain the maximum width using area under the curve functions following published methods 33 , 43 .

Additional clustering and statistical analyses were performed using custom scripts written in Python (available at GitHub ( https://github.com/anyasupp/single-neuron-synapse )). For segmentation clustering, six morphological features of FoxP2.A cells were standardized and reduced in dimensionality by projecting into principal component analysis space. The first four components, which explained 89% of the variance, were selected to use for clustering. These components were then clustered using k -means clustering with k ranging from 1 to 11. Using the elbow method, Calinski Harabasz coefficient and silhouette coefficient, we found k  = 4 to be the optimal number of k clusters.

Puncta quantification and statistics

All image files of synapse tracking experiments were blinded by an independent researcher before segmentation and puncta quantification. To count the number of FingR(PSD95)–GFP puncta, each neuron’s morphology was first segmented using the Filament function in Imaris v.8.0.2 (Bitplane). FingR(PSD95)–GFP puncta were labelled using the Spots function, thresholded using the Quality classification function at approximately 130–200 depending on the image file. The number and location of GFP puncta were also manually checked for accuracy. FingR(PSD95)–GFP puncta lying on the FingR + neuron (mKate2f red channel) were extracted using the Find Spots Close to Filament XTension add-on in IMARIS.

The percentage changes in synapse number and intensity were calculated using the following formula:

Where x represents either synapse number or intensity and x t  − 1 is the respective synapse number or intensity at the previous timepoint. Statistical tests were implemented using Python 62 . Values in the figures represent the average ± 68% CI unless stated otherwise.

Synapse intensity was calculated using the ratio of the normalized average FingR(PSD95)–GFP intensity and mKate2f, to account for depth-dependent signal reduction 63 . First, the average FingR(PSD95)–GFP and mKate2f (cell morphology) intensities at the same location within the neuron were extracted using the Imaris Spots function. Next, these average intensity values were normalized to their respective channel maximum and minimum value to account for larval position inconsistencies between imaging as follows:

Depth-dependent signal reduction was corrected by calculating the FingR(PSD95)–GFP:mKate2f ratio as follows:

Before statistical analysis, all datasets were tested for normality using the Shapiro–Wilk test followed by direct visual inspection of Q – Q plots. For repeated-measures design, the data were first tested for sphericity using Mauchly’s test; repeated-measures or mixed ANOVAs were then performed, corrected with Greenhouse–Geisser correction when sphericity was violated, followed by post hoc t -tests corrected with Benjamini–Hochberg correction for multiple comparisons. For multiple-sample comparisons, equal variances were tested using Levene’s tests. If variances were equal, either one-way ANOVA (multiple groups) with post hoc Benjamini–Hochberg correction or Student’s t -tests (two groups) were performed to test for significant differences. If variances were unequal, Kruskal–Wallis (multiple groups) with Dunn’s multiple-comparison correction or Mann–Whitney U -tests (two groups) were performed to test for significant differences. All of the statistical analyses performed are provided in Supplementary Data  1 .

per3 circadian rhythm bioluminescence assay

Larvae (6 d.p.f.) from a Tg(per3:luc ) g1 ;Tg(elavl3:EGFP ) knu3 incross were individually placed into wells of 24-well plates in water containing 0.5 mM beetle luciferin (Promega). From ZT14 (the light to dark transition) the next day, half of the larvae were subjected to a sleep deprivation paradigm (see the ‘Sleep deprivation assay’ section) under dim red light, while the others were left undisturbed in similar lighting conditions. At the end of the 4 h sleep deprivation period, the larvae were individually transferred to the wells of a white-walled 96-round-well plate (Greiner Bio-One) and sealed with an oxygen-permeable plate-seal (Applied Biosystems). Bioluminescence photon counts, reflecting luciferase expression driven by the per3 promoter, were sampled every 10 min for three consecutive days, in constant dark at 28 °C, using the TopCount NXT scintillation counter (Packard).

HCR fluorescence in situ hybridization

FoxP2.A neurons were sparsely labelled with GFP by co-electroporating wild-type AB larvae with FoxP2.A:Gal4FF and UAS:eGFP 1 at 500 ng µl −1 each (see the ‘Single-cell FingR(PSD95) expression using electroporation’ section). Whole-mount hybridization chain reaction (HCR) was performed on larvae with FoxP2.A neurons positive for GFP at 7 d.p.f. using an adapted protocol from a previous study 64 . In brief, larvae were fixed with 4% PFA and 4% sucrose overnight at 4 °C. The next day, the larvae were washed with PBS to stop fixation and the brains were removed by dissection. The dissected specimens were permeabilized using proteinase K (30 µg ml −1 ) for 20 min at room temperature, then washed twice in PBS with 0.1% Tween-20 (PBST), before being post-fixed in 4% PFA for 20 min at room temperature. The larvae were then washed in 0.1% PBST and prehybridized with prewarmed HCR hybridization buffer (Molecular Instruments) for 30 min at 37 °C.

Probes targeting multiple genes associated with different types of adenosine or adrenergic receptors were combined and labelled to the same hairpins. For example, probes detecting adora1a-b (encoding adenosine receptor A1a and A1b) contain initiators that correspond with hairpins (B3) labelled with Alexa 546 fluorophore, whereas adora2aa , adora2ab and adora2b (encoding adenosine receptors A2aa, A2ab and A2b) contain initiators that correspond with hairpins (B5) labelled with Alexa 647 fluorophore (Supplementary Data  2 ). Probe solutions consisting of cocktails of HCR probes for each transcript (Thermo Fisher Scientific) were prepared with a final concentration of 24 nM per HCR probe in HCR hybridization buffer. The larvae were then incubated in probe solutions overnight at 37 °C. Excess probes were removed by washing larvae four times for 15 min with probe wash buffer (Molecular Instruments) at 37 °C followed by two 5 min washes of 5× SSCT buffer (5× sodium chloride sodium citrate and 0.1% Tween-20) at room temperature. Preamplification was performed by incubating the samples with amplification buffer (Molecular Instruments) for 30 min at room temperature. Hairpin h1 and hairpin h2 were prepared separately by snap-cooling 4 µl of 3 µM stock at 95 °C for 20 min and 20 °C for 20 min. The larvae were then incubated with h1 and h2 hairpins in 200 µL amplification buffer overnight in the dark at room temperature. Excess hairpins were washed thoroughly the next day twice for 5 min and three times for 30 min with 5× SSCT at room temperature. The specimens were then imaged using a ×20 water-immersion objective and the LSM 980 confocal microscope with Airyscan 2 (Zeiss). The endogenous GFP signal from FoxP2.A was visualized without amplification.

Reporting summary

Further information on research design is available in the  Nature Portfolio Reporting Summary linked to this article.

Data availability

The data are available at GitHub ( https://github.com/anyasupp/single-neuron-synapse ) 65 .  Source data are provided with this paper.

Code availability

The code used to generate figures in this manuscript can be found at GitHub ( https://github.com/anyasupp/single-neuron-synapse ) 65 . The sleep analysis code is available at GitHub ( https://github.com/JRihel/Sleep-Analysis ) 66 . The frame by frame analysis code can be found at GitHub ( https://github.com/francoiskroll/FramebyFrame ) 67 .

Anafi, R. C., Kayser, M. S. & Raizen, D. M. Exploring phylogeny to find the function of sleep. Nat. Rev. Neurosci. 20 , 109–116 (2019).

Article   CAS   PubMed   Google Scholar  

Tononi, G. & Cirelli, C. Sleep and the Price of Plasticity: From Synaptic and Cellular Homeostasis to Memory Consolidation and Integration. Neuron 81 , 12–34 (2014).

Article   CAS   PubMed   PubMed Central   Google Scholar  

de Vivo, L. et al. Ultrastructural evidence for synaptic scaling across the wake/sleep cycle. Science (80-.). 355 , 507–510 (2017).

Article   PubMed Central   ADS   Google Scholar  

Maret, S., Faraguna, U., Nelson, A. B., Cirelli, C. & Tononi, G. Sleep and waking modulate spine turnover in the adolescent mouse cortex. Nat. Neurosci. 14 , 1418–1420 (2011).

Van Dongen, H. P. A., Maislin, G., Mullington, J. M. & Dinges, D. F. The cumulative cost of additional wakefulness: Dose-response effects on neurobehavioral functions and sleep physiology from chronic sleep restriction and total sleep deprivation. Sleep 26 , 117–126 (2003).

Article   PubMed   Google Scholar  

Abel, T., Havekes, R., Saletin, J. M. & Walker, M. P. Sleep, plasticity and memory from molecules to whole-brain networks. Current Biology 23 , 774–788 (2013).

Article   Google Scholar  

Tononi, G. & Cirelli, C. Sleep and synaptic homeostasis: a hypothesis. Brain Res. Bull. 62 , 143–150 (2003).

Tononi, G. & Cirelli, C. Sleep function and synaptic homeostasis. Sleep Medicine Reviews 10 , 49–62 (2006).

Noya, S. B. et al. The forebrain synaptic transcriptome is organized by clocks but its proteome is driven by sleep. Science (80-.). 366, (2019).

Gilestro, G. F., Tononi, G. & Cirelli, C. Widespread Changes in Synaptic Markers as a Function of Sleep and Wakefulness in Drosophila. Science (80-.). 324 , 109–112 (2009).

Article   CAS   PubMed Central   ADS   Google Scholar  

Vyazovskiy, V. V., Cirelli, C., Pfister-Genskow, M., Faraguna, U. & Tononi, G. Molecular and electrophysiological evidence for net synaptic potentiation in wake and depression in sleep. Nat. Neurosci. 11 , 200–208 (2008).

De Vivo, L. et al. Evidence for sleep-dependent synaptic renormalization in mouse pups. Sleep 42 , zsz184 (2019).

Article   PubMed   PubMed Central   Google Scholar  

Liu, Z. W., Faraguna, U., Cirelli, C., Tononi, G. & Gao, X. B. Direct evidence for wake-related increases and sleep-related decreases in synaptic strength in rodent cortex. J. Neurosci. 30 , 8671–8675 (2010).

Yang, G. & Gan, W. B. Sleep contributes to dendritic spine formation and elimination in the developing mouse somatosensory cortex. Dev. Neurobiol. 72 , 1391–1398 (2012).

Miyamoto, D., Marshall, W., Tononi, G. & Cirelli, C. Net decrease in spine-surface GluA1-containing AMPA receptors after post-learning sleep in the adult mouse cortex. Nat. Commun. 12 , 1–13 (2021).

Article   ADS   Google Scholar  

Appelbaum, L. et al. Circadian and homeostatic regulation of structural synaptic plasticity in hypocretin neurons. Neuron 68 , 87–98 (2010).

Cary, B. A. & Turrigiano, G. G. Stability of neocortical synapses across sleep and wake states during the critical period in rats. Elife 10 , 1–28 (2021).

Torrado Pacheco, A., Bottorff, J., Gao, Y. & Turrigiano, G. G. Sleep Promotes Downward Firing Rate Homeostasis. Neuron 109 , 530–544.e6 (2021).

Ognjanovski, N., Maruyama, D., Lashner, N., Zochowski, M. & Aton, S. J. CA1 hippocampal network activity changes during sleep-dependent memory consolidation. Front. Syst. Neurosci. 8 , 61 (2014).

Durkin, J. & Aton, S. J. Sleep-dependent potentiation in the visual system is at odds with the synaptic homeostasis hypothesis. Sleep 39 , 155–159 (2016).

Ruiz, S. et al. Rhythmic Changes in Synapse Numbers in Drosophila melanogaster Motor Terminals. PLoS One 8 , e67161 (2013).

Article   CAS   PubMed   PubMed Central   ADS   Google Scholar  

Chauvette, S., Seigneur, J. & Timofeev, I. Sleep Oscillations in the Thalamocortical System Induce Long-Term Neuronal Plasticity. Neuron 75 , 1105–1113 (2012).

Weiss, J. T. & Donlea, J. M. Sleep deprivation results in diverse patterns of synaptic scaling across the Drosophila mushroom bodies. Curr. Biol. 1–14 (2021).

Barnes, S. J. et al. Deprivation-Induced Homeostatic Spine Scaling In Vivo Is Localized to Dendritic Branches that Have Undergone Recent Spine Loss. Neuron 1–12 (2017).

Gross, G. G. et al. Recombinant probes for visualizing endogenous synaptic proteins in living neurons. Neuron 78 , 971–985 (2013).

Dempsey, W. P. et al. Regional synapse gain and loss accompany memory formation in larval zebrafish. Proc. Natl. Acad. Sci. USA 119 , e2107661119 (2022).

Son, J.-H. et al. Transgenic FingRs for Live Mapping of Synaptic Dynamics in Genetically-Defined. Neurons. Sci. Rep. 6 , 18734 (2016).

Article   CAS   PubMed   ADS   Google Scholar  

Cho, K. O., Hunt, C. A. & Kennedy, M. B. The rat brain postsynaptic density fraction contains a homolog of the drosophila discs-large tumor suppressor protein. Neuron 9 , 929–942 (1992).

Kistner, U. et al. SAP90, a rat presynaptic protein related to the product of the Drosophila tumor suppressor gene dlg- A . J. Biol. Chem. 268 , 4580–4583 (1993).

Keith, D. & El-Husseini, A. Excitation control: Balancing PSD-95 function at the synapse. Front. Mol. Neurosci. 1 , 4 (2008).

Schnell, E. et al. Direct interactions between PSD-95 and stargazin control synaptic AMPA receptor number. Proc. Natl. Acad. Sci. USA 99 , 13902–13907 (2002).

Cook, S. G., Goodell, D. J., Restrepo, S., Arnold, D. B. & Bayer, K. U. Simultaneous Live Imaging of Multiple Endogenous Proteins Reveals a Mechanism for Alzheimer’s-Related Plasticity Impairment. Cell Rep. 27 , 658–665 (2019).

Nikolaou, N. & Meyer, M. P. Lamination Speeds the Functional Development of Visual Circuits. Neuron 88 , 999–1013 (2015).

Niell, C. M., Meyer, M. P. & Smith, S. J. In vivo imaging of synapse formation on a growing dendritic arbor. Nat. Neurosci. 7 , 254–260 (2004).

Zhang, L. I., Tao, H. W., Holt, C. E., Harris, W. A. & Poo, M. M. A critical window for cooperation and competition among developing retinotectal synapses. Nature 395 , 37–44 (1998).

Gahtan, E., Tanger, P. & Baier, H. Visual prey capture in larval zebrafish is controlled by identified reticulospinal neurons downstream of the tectum. J. Neurosci. 25 , 9294–9303 (2005).

Bollmann, J. H. The Zebrafish Visual System: From Circuits to Behavior. Annu. Rev. Vis. Sci. 5 , 269–293 (2019).

Bonkowsky, J. L. et al. Domain-specific regulation of foxP2 CNS expression by lef1. BMC Dev. Biol 8 , 1–15 (2008).

Whitmore, D., Foulkes, N. S. & Sassone-Corsi, P. Light acts directly on organs and cells in culture to set the vertebrate circadian clock. Nature 404 , 87–91 (2000).

Prober, D. A., Rihel, J., Onah, A. A., Sung, R.-J. & Schier, A. F. Hypocretin/Orexin Overexpression Induces An Insomnia-Like Phenotype in Zebrafish. J. Neurosci. 26 , 13400–13410 (2006).

Wang, H. et al. Single-cell in vivo imaging of cellular circadian oscillators in zebrafish. PLoS Biol. 18 , e3000435 (2020).

Kaneko, M. & Cahill, G. M. Light-dependent development of circadian gene expression in transgenic zebrafish. PLoS Biol. 3 , e34 (2005).

Robles, E., Smith, S. J. & Baier, H. Characterization of genetically targeted neuron types in the zebrafish optic tectum. Front. Neural Circuits 5 , 1–14 (2011).

Meek, J. Functional anatomy of the tectum mesencephali of the goldfish. An explorative analysis of the functional implications of the laminar structural organization of the tectum. Brain Res. Rev. 287 , 247–297 (1983).

Article   CAS   Google Scholar  

Rihel, J. et al. Zebrafish Behavioral Profiling Links Drugs to Biological Targets and Rest/Wake Regulation. Science (80-.). 327 , 348–351 (2010).

Article   CAS   ADS   Google Scholar  

Shaw, P. J., Tortoni, G., Greenspan, R. J. & Robinson, D. F. Stress response genes protect against lethal effects of sleep deprivation in Drosophila. Nature 417 , 287–291 (2002).

Borbély, A. A. & Achermann, P. Sleep Homeostasis and Models of Sleep Regulation. J. Biol. Rhythms 557–568 (1999).

Rechtschaffen, A., Bergmann, B. M., Gilliland, M. A. & Bauer, K. Effects of method, duration, and sleep stage on rebounds from sleep deprivation in the rat. Sleep 22 , 11–31 (1999).

Hamling, K. R., Harmon, K. & Schoppik, D. The Nature and Origin of Synaptic Inputs to Vestibulospinal Neurons in the Larval Zebrafish. eNeuro 7, (2023).

Antinucci, P., Folgueira, M. & Bianco, I. H. Pretectal neurons control hunting behaviour. Elife 8 , 1–34 (2019).

Neki, D., Nakayama, H., Fujii, T., Matsui-Furusho, H. & Oda, Y. Functional motifs composed of morphologically homologous neurons repeated in the hindbrain segments. J. Neurosci. 34 , 3291–3302 (2014).

Gandhi, A. V., Mosser, E. A., Oikonomou, G. & Prober, D. A. Melatonin Is required for the circadian regulation of sleep. Neuron 85 , 1193–1199 (2015).

Sakamoto, H., Fukuda, S., Minakawa, Y. & Sawamura, S. Clonidine induces sedation through acting on the perifornical area and the locus coeruleus in rats. J. Neurosurg. Anesthesiol. 25 , 399–407 (2013).

Porkka-Heiskanen, T. et al. Adenosine: A mediator of the sleep-inducing effects of prolonged wakefulness. Science (80-.). 276 , 1265–1267 (1997).

Article   CAS   PubMed Central   Google Scholar  

Diering, G. H. et al. Homer1a drives homeostatic scaling-down of excitatory synapses during sleep. Science (80-.). 515 , 511–515 (2017).

Miyamoto, D., Hirai, D. & Murayama, M. The roles of cortical slow waves in synaptic plasticity and memory consolidation. Front. Neural Circuits 11 , 92 (2017).

Zhang, C., Gao, J., Zhang, H., Sun, L. & Peng, G. Robo2-Slit and Dcc-Netrin1 coordinate neuron axonal pathfinding within the embryonic axon tracts. J. Neurosci. 32 , 12589–12602 (2012).

Lister, J. A., Robertson, C. P., Lepage, T., Johnson, S. L. & Raible, D. W. nacre encodes a zebrafish microphthalmia-related protein that regulates neural-crest-derived pigment cell fate. Development 126 , 3757–3767 (1999).

Sheets, L., Trapani, J. G., Mo, W., Obholzer, N. & Nicolson, T. Ribeye is required for presynaptic CaV1.3a channel localization and afferent innervation of sensory hair cells. Development 1309-1319 (2011).

Singh, C., Oikonomou, G. & Prober, D. A. Norepinephrine is required to promote wakefulness and for hypocretin-induced arousal in zebrafish. Elife 4 , 1–22 (2015).

Zhdanova, I. V., Wang, S. Y., Leclair, O. U. & Danilova, N. P. Melatonin promotes sleep-like state in zebrafish. Brain Res. 903 , 263–268 (2001).

Vallat, R. Pingouin: statistics in Python. J. Open Source Software. 3, (2018).

Yoon, S. et al. Deep optical imaging within complex scattering media. Nature Reviews Physics 2 , 141–158 (2020).

Choi, H. M. T. et al. Third-generation in situ hybridization chain reaction: Multiplexed, quantitative, sensitive, versatile, robust. Development. 145 , 1–10 (2018).

Suppermpool, A., Lyons, D. G., Broom, E. & Rihel, J. Data and code for ‘Sleep pressure modulates single-neuron synapse number in zebrafish’. GitHub github.com/anyasupp/single-neuron-synapse (2024).

Rihel, J. Sleep-Analysis (GitHub, 2023); github.com/JRihel/Sleep-Analysis .

Kroll, F. FramebyFrame (GitHub, 2024); github.com/francoiskroll/FramebyFrame .

Download references

Acknowledgements

We thank all of the current and past members of the Rihel laboratory for discussions and feedback on this project and the members of the zebrafish community for sharing protocols and reagents; S. Lim for her assistance in blinding experimental files; A. Gilbert for help with early SD experiments; L. Sheets for her guidance on synapse immunohistochemistry; L. Elias for her guidance on gene expression; N. Nikolaou for sharing his knowledge of FoxP2.A neurons; C. Trivedi for his help with designing HCR probes; and the staff at the UCL Fish Facility for fish husbandry and UCL Imaging Facility for their expertise. This work was supported by UCL Research Scholarship (to A.S.), an EMBO Fellowship awarded to D.G.L. (ALTF 1097-2016), a Medical Research Council studentship (MR/W006774/1 to E.B.), a European Research Council Starting Grant (282027 to J.R.) and a Wellcome Trust Investigator Award (217150/Z/19/Z to J.R.).

Author information

• Anya Suppermpool

Present address: UCL Ear Institute, University College London, London, UK

Authors and Affiliations

Department of Cell and Developmental Biology, University College London, London, UK

Anya Suppermpool, Declan G. Lyons, Elizabeth Broom & Jason Rihel

You can also search for this author in PubMed   Google Scholar

Contributions

J.R. and A.S. conceived and designed all of the experiments with input from D.G.L. A.S. performed all of the experiments with help from D.G.L. (circadian clock/dark pulse and SD), E.B. (HCR) and J.R. (SD). A.S., D.G.L. and J.R. wrote the manuscript with input from E.B.

Corresponding author

Correspondence to Jason Rihel .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Peer review

Peer review information.

Nature thanks Chiara Cirelli, Jiu-Lin Du, David Schoppik, William Wisden and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data figures and tables

Extended data fig. 1 the modified fingr(psd95)-gfp construct labels synapses in vivo..

a-a” , Maximum projection (Z-stack, ~10 μm) of anti-MAGUK immunohistochemistry and endogenous fluorescence of FingR(PSD95)-GFP in the spinal cord of 2 dpf Tg(mnx1:Gal4 ) larvae. Examples of FingR(PSD95)+ puncta co-labelled by anti-MAGUK are indicated by white arrowheads; an example of a FingR(PSD95)+ not labelled by anti-MAGUK is indicated by the blue arrowhead. b-b”’ , Higher magnification (white box from a ) depicting how sectional grey values for each synapse were obtained. b , The FingR(PSD95)-GFP channel showing part of a neuron with its nucleus (asterisk) and synaptic puncta (green). Dotted lines indicate example cross-sectional areas obtained for each synapse. b ’, Anti-MAGUK puncta of the same neuron. b” , b”’ , FingR(PSD95)-GFP and MAGUK channels merged, with examples of cross-sections 1–4. c , Examples of normalized cross-sectional grey values for anti-MAGUK signals and FingR(PSD95)-GFP signal for the same puncta (numbered 1–4 in b”’ ). Three examples in which FingR(PSD-95)-GFP co-localized with anti-MAGUK signals (#1–3) and one example (#4) where a FingR(PSD-95)-GFP punctum did not co-localize with MAGUK. See  Methods for details. d , Percentage of FingR(PSD-95)-GFP synapses that co-localized with anti-MAGUK+ puncta (blue). As a control for chance co-localization, the calculation was repeated on images in which the anti-MAGUK image was rotated by 90° relative to the FingR(PSD-95)-GFP channel. ****P = 1.1 × 10 −83 Chi-square. e , Histogram of the distance between all co-localized FingR(PSD95)-GFP and anti-MAGUK cross-sectional grey value peaks. f - g , The intensity and Full Width Half Max (FWHM) of FingR(PSD95)-GFP and anti-MAGUK puncta are weakly, but significantly, positively correlated. Blue and red lines depict the linear regression curve and 95% CI for the colocalized and non-colocalized populations, respectively. n = 540 puncta, 5 fish (data as in d ). h , Percentage of anti-MAGUK+ puncta that co-localized with FingR(PSD-95)-GFP synapses (blue). As a control for chance co-localization, the calculation was repeated on images in which the FingR(PSD-95)-GFP image was rotated by 90° relative to the anti-MAGUK channel. ****P = 3.1 × 10 −14 Chi-square. i , Histogram of the distance between co-localized anti-MAGUK and FingR(PSD95)-GFP cross-sectional grey value peaks. Scale bar: 5 μm ( a-b”’ ).

Extended Data Fig. 2 The synapse number of single tectal neurons is developmentally stable at 6–9 dpf.

a , The full map of synapse tracking from the neuron in Fig. 1c . Each column depicts a synapse, and the colour indicates the normalized GFP intensity of each synapse. In this example, 56 synapses disappeared and 20 synapses appeared during the imaging, resulting in a net change of −36 synapses. Grey bars depict night (ZT14-24). b , Example of a single FoxP2.A:FingR(PSD95)+ neuron imaged through development from 4–10 dpf. Nuclei and synapses are FingR(PSD95)-GFP+ (green), and cellular morphology is labelled by mKate2f (magenta). White arrowheads indicate examples of puncta that persisted through time. Blue arrowheads indicate examples of synapses gained/lost through time. c , Synapse counts across all neurons (average and 68% CI) ( left ) and for single neurons through 4–10 dpf ( right ). d , Average percentage change in synapse number and 68% CI calculated from the previous time point ( left ) and for each neuron ( right ). The percentage change in synapse number across time is close to zero between 6–9 dpf. n = 5 cells, 5 larvae. Scale bar: 15 μm ( b ).

Extended Data Fig. 3 Example of a single FoxP2.A:FingR(PSD95)+ neuron at ZT14 and ZT18.

a , A single FoxP2.A:FingR(PSD95)+ tectal neuron imaged at ZT14 and ZT18. Nuclei and synapses are FingR(PSD95)-GFP+ (green), and cellular morphology is labelled by mKate2f (magenta). b , Higher magnification of the primary dendrite segment (white box in a ). Right panels show semi-automatic skeletonization (lines) of neurites and detection of FingR(PSD95)-GFP puncta (grey spheres, Methods ). c , Higher magnification of a section of the distal arbour (white box in a ). FingR(PSD95)-GFP+ puncta that appeared (blue circles and arrowheads) and disappeared (yellow circles and arrowheads) between ZT14 and ZT18 can be observed. d , Schematic showing imaging times (black arrows) at ZT14 and ZT18 on the night of 7 dpf. Scale bars: 10 μm ( a ) and 2.5 μm ( b , c ).

Extended Data Fig. 4 Extended tracking of single neurons over multiple days.

a , Larvae were raised on 14h–10h LD cycles (blue), on constant light (pink), or switched from LD to LL at 6 dpf (‘free running’, FR, green) and repeatedly imaged (arrows) at ZT0 and ZT10 for each day from 7–9 dpf. b - c , The average (68%CI) ( b ) and percentage change ( c ) for synapse counts at each timepoint in LD (blue), LL (pink), or FR (green) conditions from 7–9 dpf ( left ). Each n = neuron is plotted as a single line ( right ). d - e , Average synapse counts and percentage change (68%CI) for ZT0 and ZT10 combined across all tracked days for each lighting condition (LD, 13 independent experiments; LL, 4 experiments, and FR, 4 experiments). The ZT10 timepoint from 9 dpf was excluded to avoid interference from a new developmental round of synaptogenesis. f , Schematic of experiment to test whether repeated imaging affected synapse number and strength measurements. Larvae raised in LD (indicated by white and grey boxes) were either imaged six times between 7–9 dpf at ZT0 and ZT10 (Tracked, orange) or imaged at ZT0 on 7 dpf and ZT10 on 9 dpf (Control, green). g - h , Average (with 68%CI) synapse counts ( g ) and normalized average synapse intensity ( h ) at the first and last time point (7 dpf ZT0 and 9 dpf ZT10) for tracked and control larvae ( left ). The percentage changes in synapse number ( g , right ) and average synapse intensity ( h , right ) were not statistically different between tracked and control larvae. Controls: n = 6 neurons, 4 larvae; Tracked: n = 14 neurons, 14 larvae collected over 8 independent experiments. ns, P > 0.05 Student’s t-test, two tailed.

Extended Data Fig. 5 FoxP2.A tectal neurons have four morphological subtypes.

a , Principal component analysis using the subtype morphological features depicted in Fig. 2a . Four principal components (dotted line) account for >85% of the variance. b , The optimal number of clusters for k-means clustering was determined using the elbow method by plotting the within-cluster sum of squares. Four clusters were chosen (dotted line). c , The six features used to cluster FoxP2.A neurons (collected over 26 experiments) by morphological subtype. Boxes depict the median and interquartile range and the whiskers represent the distribution for each parameter. The slashed zero means the feature is absent. d - f ( left ), Synapse counts with 68%CI ( d ), average change (68%CI) in synapse counts ( e ), and percentage change (68%CI) in synapse counts ( f ) in different FoxP2.A tectal neuron subtypes of larvae raised in normal LD conditions. d - f ( right ), Each neuron is plotted, grouped by subtype. g , Average (68%CI) synapse counts of tectal subtypes ( left ) and for each n= neuron ( right ) across multiple days under clock-break (LL) conditions. Note the lack of Type 2 neurons in LL. h , Average (68%CI) synapse counts during the subjective day or night under clock-break conditions. i , Average change (68%CI) in synapse counts ( left ) and single neurons ( right ) across multiple days under clock-break conditions, sorted by tectal subtype. j , the average change (68%CI) in synapse counts for the subjective day and night under clock-break conditions. Data in g - j are from 4 independent experiments.

Extended Data Fig. 6 FingR(PSD95):GFP signal intensity increases during the day and decreases at night in some, but not all tectal subtypes.

a , Average and 68% CI of normalized synapse intensity on LD, LL, and FR conditions across one day and night for a subset of tectal neurons from Fig. 2 imaged under identical microscopy settings to enable intensity measurements. Note that the loss of the circadian clock alters the relative abundance of Type 1 and Type 2 neurons. b , Percentage change (mean and 68% CI) in normalized synapse intensity calculated as in Fig. 1 . Compared to Type 2 neurons, Type 3 (p = 0.026; g = 1.777) and Type 4 (p = 0.026; g = 1.651) neurons have increased synapse intensities during the day (mixed ANOVA, interaction (subtype*time) p = 0.03, post-hoc Benjamini-Hochberg, one tailed). c , Both Type 3 (p = 0.026; g = 1.691) and Type 4 (p = 0.026; g = 1.408) neurons have significantly increased synapse intensities (with 68%CI) during the day relative to clock-break (LL) conditions (mixed ANOVA, interaction (condition*time) p = 0.006, post-hoc Benjamini-Hochberg, one tailed). Data are collected from 8 independent LD, 4 LL, and 4 FR experiments.

Extended Data Fig. 7 Tectal subtype labelling does not bias larval sleep amount and sleep-wake states have non-uniform effects on synapses within neuronal compartments.

a , Schematic of behavioural and synapse tracking experiment set up. Larval locomotor behaviour was tracked on a 14 h–10 h LD cycle from 6–8 dpf. The average activity ( ± 95% CI) of 10 example larvae are plotted across two days and nights. Larvae were removed from the tracking arena and imaged at lights on (ZT0) and again at ZT10 (dotted red bars). White and grey boxes indicate day and night periods, respectively. b , 7 dpf Larvae had similar levels of sleep and sleep bout lengths at night ( ± SEM) regardless of the FoxP2.A tectal neurons subtype labelled in each larva (ns, p > 0.05, Kruskal-Wallis; 5 independent experiments). c , For each neuron/larva, the average percentage change of synapse number is plotted versus the average 7 dpf night-time sleep. d , Type 2 tectal neurons were divided into four segments: the primary neurite, proximal arbour, inter-arbour area, and distal arbour. e , The average and 68% CI of synapse number and intensity dynamics within each of the four segments. Grey lines represent segments from individual neurons. *P = 0.037, repeated-measures ANOVA with Greenhouse-Geisser correction. f , Proximal and distal arbours synapse number dynamics are not correlated. The relationship between the absolute and relative (%) synapse number change of the proximal and distal arbours of individual Type 2 neurons during the day and night phase. Linear regressions in c and f are fitted with 95% CI.

Extended Data Fig. 8 Sleep deprivation affects synapse number in tectal neuron subtypes.

a , Percentage change of total sleep ( left ) and average sleep bout length ( right ) of each larva (dots) in the 6 hr post SD (ZT18-24, 7dpf), normalized to the circadian-matched time at 6 dpf. The black lines depict the average ± SEM. *P < 0.02, one-way ANOVA. b , The SD method did not alter circadian clock phase as measured by the bioluminescence driven by a Tg(per3-luc ) reporter line for the clock gene per3 expression. The detrended per3 bioluminescence rhythms ( ± 95%CI) remained in phase for both SD (n = 14 larvae) and control (n = 12) larvae over multiple days of constant dark conditions. Circadian time (CT = 0 last lights ON transition). c , The percentage change in synapse number within each neuron between imaging sessions at ZT14 and ZT18, and between imaging at ZT18 and ZT24. d , Average (68%CI) for net synapse change per hour for FoxP2.A tectal subtypes in control or sleep deprived larvae. Type 3, but not Type 4 neurons significantly gain synapses after SD (Mixed ANOVA, post-hoc Benjamini-Hochberg, one tailed **p = 0.01, g = 1.266) and subsequently lose them (p = 0.014, g = −1.034) relative to controls. Type 2 lacks enough matched controls to assess. e , Sleep amount for early and late sleepers in the early (ZT14-18) and late (ZT18-24) phase of the night (5 independent experiments). The black lines depict the average ± SEM. f , For each neuron/larva, changes in synapse number during extended wakefulness did not correlate with either the subsequent total sleep or average sleep bout lengths (mean ± 95% CI). g , Changes in synapse numbers for each neuron/larva did not significantly correlate with the average sleep bout lengths during the early and late night of controls, or after SD (mean ± 95% CI). *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001, Mixed ANOVA interaction (condition*time), post-hoc Benjamini-Hochberg, two tailed.

Extended Data Fig. 9 Examples of manipulated single FoxP2.A:FingR(PSD95)+ neurons and clonidine and evidence that daytime drug treatment reduced sleep the following night.

a , left Example FoxP2.A:FingR(PSD95)+ tectal neurons imaged before (ZT14), immediately after (ZT18), and 6 h after (ZT24) sleep deprivation and control. Nuclei and synapses are FingR(PSD95)-GFP+ (green), and cellular morphology is labelled by mKate2f (magenta). Right , Higher magnification (dotted white box) showing the same dendritic segments at each time point, with examples of synapses lost (yellow arrows and dotted circles) or gained (blue arrows and circles). Note that, for illustrative purposes, the dendrites are depicted at a different angle in these higher magnification images. b , An example neuron before (ZT5) or after (ZT10) exposure to clonidine and 2-chloroadenosine. Scale bars: 15 μm ( a , b left ) and 5 μm ( a , b right ). c , Larvae (n = 80) exposed to lights OFF at mid-day (ZT8, first arrow in schematic) took longer to sleep (mean ± SEM) compared to lights OFF at the end of day (ZT14, 2nd arrow). ****P = 2.27 × 10 −15 , Kruskal-Wallis. d , Average locomotor activity ( ± 95%CI) on a 14 hr:10 hr LD cycle before, during, and after a 5 hr midday (ZT5-10, 7 dpf, shaded purple panel) exposure to melatonin (n = 31 larvae), clonidine (n = 32), or DMSO (n = 32). Data from two independent experiments. e , Larvae treated with either melatonin or clonidine from ZT5-10 had reduced and delayed sleep ( ± SEM) in first hour of the night (ZT14-15) compared to controls. *P < 0.05, **P < 0.01, ****P < 0.0001 Dunnett’s Test.

Extended Data Fig. 10 Drug-evoked day time sleep induces synapse loss only when clonidine and 2-chloroadenosine are co-administered.

a - b , Clonidine-, 2-chloroadenosine-, and/or melatonin-treated larvae have a lower average activity ( ± SEM) and longer average sleep bout lengths ( ± SEM) during the 5 hr drug period compared to DMSO treated controls. c , The average percentage change in synapse number ( ± SEM) within each neuron of DMSO, clonidine-, 2-chloroadenosine-, and/or melatonin-treated larvae. *P < 0.05, **P < 0.01, ****P < 0.0001 Kruskal-Wallis with post-hoc Dunn’s test ( b left and right; and c , left) or one-way ANOVA ( a right, c right). d , The average activity of larvae before, during and after treatment with either 30 µM clonidine or DMSO from ZT5-10 (purple shaded area) at 7 dpf. 1-minute dark pulses were given every 30 min during the treatment period to test for responsiveness. e , Higher resolution time-course of average locomotor activity during the drug treatment and dark-pulse period (ZT5-10). f , Both clonidine and DMSO-treated larvae respond to dark pulse with an increase in locomotion, known as the visuomotor response or dark photokinesis. Shown is the average locomotor response to a single 1-minute dark pulse delivered at ZT7. g , Locomotor activity for each larva-treated with clonidine (1-minute bin) at the time of dark pulse (ZT7) shown in d . Of the 13 larvae that were inactive at the onset of the 1-minute dark pulse, 12 rapidly increased their locomotor activity within 1 min.

Extended Data Fig. 11 FoxP2.A+ neurons express adenosine and adrenergic receptors transcripts.

Examples of adrenergic and adenosine receptor transcripts that colocalize with labelled FoxP2.A+ neurons (middle and right panel) as detected by in situ Hybridization Chain Reaction (HCR, see  Methods ). a , A single labelled tectal neuron (green) colocalizes with a cocktail of HCR probes that detect adora1a-b (yellow, encoding for adenosine receptors A1a and A1b) and adora2aa , -ab , -b (magenta, encoding for adenosine receptors A2aa, A2ab, and A2b) transcripts. b , Single FoxP2.A+ neuron (green) also colocalize with an HCR probe cocktail that detects adra1 aa , -ab , -ba , -bb , -d (yellow, encoding zebrafish α1 adrenergic receptor orthologs) and adra2a , -c , -da (magenta, encoding zebrafish α2 adrenergic receptor orthologs) transcripts. Scale bar: 10 μm ( a , b ). Representative data from 5 larvae. Images of co-localized transcripts chosen from n = 11 neurons ( a ) and n = 10 neurons ( b ).

Supplementary information

Reporting summary, peer review file, supplementary data 1.

A list of all statistical tests performed in this study, with associated figure panels.

Supplementary Data 2

A list of probes for HCR analysis.

Supplementary Video 1

Example of gentle handling SD. Larvae in individual wells were manually kept awake with a paintbrush for 4 h under red light at the beginning of the night (ZT14–18; Methods). Note that many, if not most, interventions did not require physically touching the animal.

Source data

Source data figs. 1–4 and source data extended data figs. 1, 2 and 4–10, rights and permissions.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Suppermpool, A., Lyons, D.G., Broom, E. et al. Sleep pressure modulates single-neuron synapse number in zebrafish. Nature (2024). https://doi.org/10.1038/s41586-024-07367-3

Download citation

Received : 14 November 2022

Accepted : 27 March 2024

Published : 01 May 2024

DOI : https://doi.org/10.1038/s41586-024-07367-3

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

This article is cited by

• Declan G. Lyons
• Jason Rihel

Nature (2024)

By submitting a comment you agree to abide by our Terms and Community Guidelines . If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Quick links

• Explore articles by subject
• Guide to authors
• Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

• Read Today's Paper

Your sleeping position could be shortening your lifespan

Nightmare fuel

Don't miss out on the headlines from Lifestyle. Followed categories will be added to My News.

According to experts, your sleeping habits and positions could be causing – or exacerbating – a range of health issues. Here’s what you need to know.

As a long-time side-sleeper, I was recently asked if I favour a certain direction by a skin therapist. She’d noticed slightly more congestion on one side of my face, a clear indication that I fall asleep, stay asleep and wake up without switching it up on my pillow. 

It’s no secret our go-to sleeping positions say a lot about us . Where you drift off to sleep in a foetal position, a starfish or with the help of a big (or little) spoon, each and every one of us rests differently . 

But as it turns out, your sleeping habits can impact far more than just your skin, according to experts. 

As several studies have shown, back pain from an uncomfortable slumber is just the tip of the iceberg when it comes to sleep posture. 

“Sleep positions can significantly affect your overall health, comfort, and the quality of your sleep,” explains physiotherapist and sleep expert Sammy Margo. “Each position has its pros and cons and understanding them can help you make adjustments for better sleep and health outcomes.”

“What you do in the day generally triggers the pain and discomfort that is felt when you sleep in certain positions,” adds sleep scientist Dr Kat Lederle. “One of the most common contributing factors to this is a sedentary lifestyle, so it is important to move regularly during the day.”

Here’s are the health implications of certain sleep positions

Front sleepers.

Though there are certain pros to sleeping on you stomach – like reduced snoring and more open airways – it’s also the position most commonly linked to neck and back pain.

“Twisting your neck to the side puts strain on your neck, and stomach sleeping can also arch your spine,” explains Margo. “Direct pressure on the face can contribute to wrinkles over time.”

If you’re a front sleeper, experts advise the use of a thin pillow (or no pillow at all) to alleviate postural pain, and to place a pillow under the pelvis to help support the lower back.

Back sleepers

Anyone who lives with a snoring partner knows back sleeping can be a little tricky to live with – for anyone in the bed. 

According to Lerderle, one of the most common (and problematic) conditions associated with back sleepers is sleep apnoea. This is where the soft tissue at the back of the throat relaxes during sleep, collapsing the airway and causing snoring or interrupted breathing. 

As the expert says, sleep apnoea disrupts the continuity and quality of a person’s sleep, leading to tiredness, which can impact daily activities like work or driving. 

“There are also physical health implications. We know that poor-quality sleep raises the risk of diabetes, heart disease and other comorbidities. Sleep apnoea opens the door to all these other conditions,” she says.

To counteract these issues, Margo suggests back sleepers rest in an elevated position, especially if chronic back or neck pain limits you from resorting to alternative positions, “The optimal position for spine alignment is lying on your back with a pillow under the knees to soften the back. This position preserves the natural contours of your spine. It can also minimise wrinkles.”

Side sleepers

Despite being the most popular sleeping position there is, experts say people could experience different health issues depending on whether they prefer the left or right side. 

For pregnant women and people who suffer from acid reflux specifically, experts warn against sleeping on the right side, “This is because the stomach is lower than your oesophagus,” Margo says.

For those with heart conditions, however, the right side is the recommended sleep position, alleviating pressure on the heart due to gravity. 

More Coverage

 “Women with hourglass figures sleeping on a soft mattress will sink into a banana shape and that will cause a strain on the spine and hips, while men who side-sleep can tend to get more pain in their shoulders as they get older and their muscles weaken,” Margo says, adding that side sleeping, in general, can also cause wrinkles and breast sagging.

According to a 2022 study by Beijing Forestry University and Chenzhou Vocational Technical College examining how sleep quality is influenced by position, subjects without sleep disorders who prefer to sleep on their side will sleep better than those who like to sleep on their back.

The sleep experts recommend side sleepers use a thick pillow for optimum neck and spine alignment and place a pillow between your knees to support the hips and lessen lower back strain.

Originally published as Your sleeping position could be shortening your lifespan

I took a road trip to the ‘real’ Ireland

On our trip through Ireland, I discovered my mum still has a heart of emerald green.

Mental health conditions in Australia linked to childhood maltreatment

It’s a harrowing finding

Help | Advanced Search

Computer Science > Machine Learning

Title: position: leverage foundational models for black-box optimization.

Abstract: Undeniably, Large Language Models (LLMs) have stirred an extraordinary wave of innovation in the machine learning research domain, resulting in substantial impact across diverse fields such as reinforcement learning, robotics, and computer vision. Their incorporation has been rapid and transformative, marking a significant paradigm shift in the field of machine learning research. However, the field of experimental design, grounded on black-box optimization, has been much less affected by such a paradigm shift, even though integrating LLMs with optimization presents a unique landscape ripe for exploration. In this position paper, we frame the field of black-box optimization around sequence-based foundation models and organize their relationship with previous literature. We discuss the most promising ways foundational language models can revolutionize optimization, which include harnessing the vast wealth of information encapsulated in free-form text to enrich task comprehension, utilizing highly flexible sequence models such as Transformers to engineer superior optimization strategies, and enhancing performance prediction over previously unseen search spaces.

Submission history

Access paper:.

• Other Formats

References & Citations

• Google Scholar
• Semantic Scholar

BibTeX formatted citation

Bibliographic and Citation Tools

Code, data and media associated with this article, recommenders and search tools.

• Institution

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs .