Modelling and exploring human sleep with event history analysis

The renin secretion profile under the influence of sleep deprivation and the neuropeptides CRH and GHRH

It is already known that during normal sleep plasma renin activity (PRA) shows oscillations with decreases during rapid-eye-movement (REM) sleep and increases during non-REM (NREM) sleep. We also know that renin correlates positively with slow-wave sleep (SWS). Sleep deprivation is known to enhance significantly SWS and slow wave activity (SWA, known as δ power). Based on these findings we addressed the question whether and to which extent sleep deprivation may affect the synchronization found between PRA and REM sleep during normal sleep and whether this synchronization is affected by other sleep regulating factors. To investigate these questions we compared sleep EEG and sleep-related free renin levels in 48 normal women and men 19-69 years old between nights before and after 40 h of sleep deprivation. During the recovery night, four bolus injections of either GHRH, CRH or placebo were injected via long catheter around sleep onset. When compared to baseline after each of the treatments SWS, SWA and renin levels increased. The characteristical oscillation profiles of renin during normal sleep were also preserved after sleep deprivation. Similar to normal sleep our data support also a distinct link between nocturnal renin secretion and SWS after sleep deprivation and that independent of the applied treatments.

General tests of the Markov property in multi-state models

Multi-state models for event history analysis most commonly assume the process is Markov. This article considers tests of the Markov assumption that are applicable to general multi-state models. Two approaches using existing methodology are considered; a simple method based on including time of entry into each state as a covariate in Cox models for the transition intensities and a method involving detecting a shared frailty through a stratified Commenges–Andersen test. In addition, using the principle that under a Markov process the future rate of transitions of the process at times $t > s$ should not be influenced by the state occupied at time $s$, a new class of general tests is developed by considering summaries from families of log-rank statistics where patients are grouped by the state occupied at varying initial time $s$. An extended form of the test applicable to models that are Markov conditional on observed covariates is also derived. The null distribution of the proposed test statistics are approximated by using wild bootstrap sampling. The approaches are compared in simulation and applied to a dataset on sleeping behavior. The most powerful test depends on the particular departure from a Markov process, although the Cox-based method maintained good power in a wide range of scenarios. The proposed class of log-rank statistic based tests are most useful in situations where the non-Markov behavior does not persist, or is not uniform in nature across patient time.

Mathematical modeling of sleep state dynamics in a rodent model of shift work

Millions of people worldwide are required to work when their physiology is tuned for sleep. By forcing wakefulness out of the body’s normal schedule, shift workers face numerous adverse health consequences, including gastrointestinal problems, sleep problems, and higher rates of some diseases, including cancers. Recent studies have developed protocols to simulate shift work in rodents with the intention of assessing the effects of night-shift work on subsequent sleep (Grønli et al., 2017). These studies have already provided important contributions to the understanding of the metabolic consequences of shift work (Arble et al., 2015; Marti et al., 2016; Opperhuizen et al., 2015) and sleep-wake-specific impacts of night-shift work (Grønli et al., 2017). However, our understanding of the causal mechanisms underlying night-shift-related sleep disturbances is limited. In order to advance toward a mechanistic understanding of sleep disruption in shift work, we model these data with two different approaches. First we apply a simple homeostatic model to quantify differences in the rates at which sleep need, as measured by slow wave activity during slow wave sleep (SWS) rises and falls. Second, we develop a simple and novel mathematical model of rodent sleep and use it to investigate the timing of sleep in a simulated shift work protocol (Grønli et al., 2017). This mathematical framework includes the circadian and homeostatic processes of the two-process model, but additionally incorporates a stochastic process to model the polyphasic nature of rodent sleep. By changing only the time at which the rodents are forced to be awake, the model reproduces some key experimental results from the previous study, including correct proportions of time spent in each stage of sleep as a function of circadian time and the differences in total wake time and SWS bout durations in the rodents representing night-shift workers and those representing day-shift workers. Importantly, the model allows for deeper insight into circadian and homeostatic influences on sleep timing, as it demonstrates that the differences in SWS bout duration between rodents in the two shifts is largely a circadian effect. Our study shows the importance of mathematical modeling in uncovering mechanisms behind shift work sleep disturbances and it begins to lay a foundation for future mathematical modeling of sleep in rodents.

•

Millions of people worldwide are required to work when their physiology is tuned for sleep.

•

Enforcing wakefulness during this time leads to numerous adverse health consequences including sleep problems and higher rates of some diseases.

•

Rodent models of shift work have illuminated some of the effects of night shift work on subsequent sleep.

•

This study uses mathematical modeling to accurately simulate rodent sleep during baseline and shift work conditions.

•

A simple mathematical framework can help us understand possible mechanisms underlying the sleep disturbances seen in shift work.

Significance of the zero sum principle for circadian, homeostatic and allostatic regulation of sleep-wake state in the rat

Sleep-wake behavior exhibits diurnal rhythmicity, rebound responses to acute total sleep deprivation (TSD), and attenuated rebounds following chronic sleep restriction (CSR). We investigated how these long-term patterns of behavior emerge from stochastic short-term dynamics of state transition. Male Sprague-Dawley rats were subjected to TSD (1day×24h, N=9), or CSR (10days×18h TSD, N=7) using a rodent walking-wheel apparatus. One baseline day and one recovery day following TSD and CSR were analyzed. The implications of the zero sum principle were evaluated using a Markov model of sleep-wake state transition. Wake bout duration (a combined function of the probability of wake maintenance and proportional representations of brief and long wake) was a key variable mediating the baseline diurnal rhythms and post-TSD responses of all three states, and the attenuation of the post-CSR rebounds. Post-NREM state transition trajectory was an important factor in REM rebounds. The zero sum constraint ensures that a change in any transition probability always affects bout frequency and cumulative time of at least two, and usually all three, of wakefulness, NREM and REM. Neural mechanisms controlling wake maintenance may play a pivotal role in regulation and dysregulation of all three states.

Modeling sleep fragmentation in sleep hypnograms: An instance of fast, scalable discrete-state, discrete-time analyses

Methods are introduced for the analysis of large sets of sleep study data (hypnograms) using a 5-state 20-transition-type structure defined by the American Academy of Sleep Medicine. Application of these methods to the hypnograms of 5598 subjects from the Sleep Heart Health Study provide: the first analysis of sleep hypnogram data of such size and complexity in a community cohort with a range of sleep-disordered breathing severity; introduce a novel approach to compare 5-state (20-transition-type) to 3-state (6-transition-type) sleep structures to assess information loss from combining sleep state categories; extend current approaches of multivariate survival data analysis to clustered, recurrent event discrete-state discrete-time processes; and provide scalable solutions for data analyses required by the case study. The analysis provides detailed new insights into the association between sleep-disordered breathing and sleep architecture. The example data and both R and SAS code are included in online supplementary materials.

Has adult sleep duration declined over the last 50+ years?

The common assumption that population sleep duration has declined in the past few decades has not been supported by recent reviews, which have been limited to self-reported data. The aim of this review was to assess whether there has been a reduction in objectively recorded sleep duration over the last 50+ years. The literature was searched for studies published from 1960 to 2013, which assessed objective sleep duration (total sleep time (TST)) in healthy normal-sleeping adults. The search found 168 studies that met inclusion criteria, with 257 data points representing 6052 individuals ages 18-88 y. Data were assessed by comparing the regression lines of age vs. TST in studies conducted between 1960 and 1989 vs. 1990-2013. Weighted regression analyses assessed the association of year of study with age-adjusted TST across all data points. Regression analyses also assessed the association of year of study with TST separately for 10-y age categories (e.g., ages 18-27 y), and separately for polysomnographic and actigraphic data, and for studies involving a fixed sleep schedule and participants' customary sleep schedules. Analyses revealed no significant association of sleep duration with study year. The results are consistent with recent reviews of subjective data, which have challenged the notion of a modern epidemic of insufficient sleep.

Sleep continuity: a new metric to quantify disrupted hypnograms in non-sedated intensive care unit patients

INTRODUCTION

Sleep in intensive care unit (ICU) patients is severely altered. In a large proportion of critically ill patients, conventional sleep electroencephalogram (EEG) patterns are replaced by atypical sleep. On the other hand, some non-sedated patients can display usual sleep EEG patterns. In the latter, sleep is highly fragmented and disrupted and conventional rules may not be optimal. We sought to determine whether sleep continuity could be a useful metric to quantify the amount of sleep with recuperative function in critically ill patients with usual sleep EEG features.

METHODS

We retrospectively reanalyzed polysomnographies recorded in non-sedated critically ill patients requiring non-invasive ventilation (NIV) for acute hypercapnic respiratory failure. Using conventional rules, we built two-state hypnograms (sleep and wake) and identified all sleep episodes. The percentage of time spent in sleep bouts (<10 minutes), short naps (>10 and <30 minutes) and long naps (>30 minutes) was used to describe sleep continuity. In a first study, we compared these measures regarding good (NIV success) or poor outcome (NIV failure). In a second study performed on a different patient group, we compared these measurements during NIV and during spontaneous breathing.

RESULTS

While fragmentation indices were similar in the two groups, the percentage of total sleep time spent in short naps was higher and the percentage of sleep time spent in sleep bouts was lower in patients with successful NIV. The percentage of total sleep time spent in long naps was higher and the percentage of sleep time spent in sleep bouts was lower during NIV than during spontaneous breathing; the level of reproducibility of sleep continuity measures between scorers was high.

CONCLUSIONS

Sleep continuity measurements could constitute a clinically relevant and reproducible assessment of sleep disruption in non-sedated ICU patients with usual sleep EEG.

Chronic escitalopram treatment attenuated the accelerated rapid eye movement sleep transitions after selective rapid eye movement sleep deprivation: a model-based analysis using Markov chains

Background

Shortened rapid eye movement (REM) sleep latency and increased REM sleep amount are presumed biological markers of depression. These sleep alterations are also observable in several animal models of depression as well as during the rebound sleep after selective REM sleep deprivation (RD). Furthermore, REM sleep fragmentation is typically associated with stress procedures and anxiety. The selective serotonin reuptake inhibitor (SSRI) antidepressants reduce REM sleep time and increase REM latency after acute dosing in normal condition and even during REM rebound following RD. However, their therapeutic outcome evolves only after weeks of treatment, and the effects of chronic treatment in REM-deprived animals have not been studied yet.

Results

Chronic escitalopram- (10 mg/kg/day, osmotic minipump for 24 days) or vehicle-treated rats were subjected to a 3-day-long RD on day 21 using the flower pot procedure or kept in home cage. On day 24, fronto-parietal electroencephalogram, electromyogram and motility were recorded in the first 2 h of the passive phase. The observed sleep patterns were characterized applying standard sleep metrics, by modelling the transitions between sleep phases using Markov chains and by spectral analysis.

Based on Markov chain analysis, chronic escitalopram treatment attenuated the REM sleep fragmentation [accelerated transition rates between REM and non-REM (NREM) stages, decreased REM sleep residence time between two transitions] during the rebound sleep. Additionally, the antidepressant avoided the frequent awakenings during the first 30 min of recovery period. The spectral analysis showed that the SSRI prevented the RD-caused elevation in theta (5–9 Hz) power during slow-wave sleep. Conversely, based on the aggregate sleep metrics, escitalopram had only moderate effects and it did not significantly attenuate the REM rebound after RD.

Conclusion

In conclusion, chronic SSRI treatment is capable of reducing several effects on sleep which might be the consequence of the sub-chronic stress caused by the flower pot method. These data might support the antidepressant activity of SSRIs, and may allude that investigating the rebound period following the flower pot protocol could be useful to detect antidepressant drug response. Markov analysis is a suitable method to study the sleep pattern.

Statistical properties of sleep-wake behavior in the rat and their relation to circadian and ultradian phases

STUDY OBJECTIVES

To examine the statistical characteristics of short-term sleep-wake architecture and to evaluate their dependence on ultradian and circadian phase.

DESIGN

Observational, time series.

SETTING

Laboratory.

PARTICIPANTS

Ten male adult Sprague-Dawley rats.

INTERVENTIONS

N/A.

MEASUREMENTS AND RESULTS

States of wakefulness (WAKE), rapid eye movement sleep (REM) and nonrapid eye movement sleep (NREM) were recorded in 5-sec epochs over 7 consecutive days. State bout durations were analyzed using parametric regression of survival curves, comparing exponential, biexponential, and power law models. WAKE survival curves were best fit by biexponential models, suggesting that there are two statistically distinct stochastic mechanisms generating two types of WAKE--"brief" WAKE and "long" WAKE. Exponential time constants varied as a function of circadian and ultradian phase, with "long" WAKE showing the largest effect. NREM survival curves exhibited biexponential and monoexponential distributions in light and dark, respectively, with weak effects of ultradian phase. REM survival curves approximated a monoexponential distribution that varied with circadian but not ultradian phase. χ(2) analysis was used in a three-state Markov model to evaluate whether conditional state transition probabilities exhibit the property of first-order dependence. This was partially confirmed, but only after accounting for heterogeneity associated with circadian and ultradian phase. However, there was evidence of residual second-order dependence indicating that additional sources of statistical heterogeneity may remain to be identified.

CONCLUSIONS

Sleep-wake state is regulated over short timescales by stochastic mechanisms. When the major sources of heterogeneity are taken into account, including two-component WAKE and NREM states, the sleep-wake system of the rat behaves, to a reasonable approximation, as a Markovian system that is modulated over ultradian and circadian timescales.

Combining hidden Markov models for comparing the dynamics of multiple sleep electroencephalograms

In this manuscript, we consider methods for the analysis of populations of electroencephalogram signals during sleep for the study of sleep disorders using hidden Markov models (HMMs). Notably, we propose an easily implemented method for simultaneously modeling multiple time series that involve large amounts of data. We apply these methods to study sleep-disordered breathing (SDB) in the Sleep Heart Health Study (SHHS), a landmark study of SDB and cardiovascular consequences. We use the entire, longitudinally collected, SHHS cohort to develop HMM population parameters, which we then apply to obtain subject-specific Markovian predictions. From these predictions, we create several indices of interest, such as transition frequencies between latent states. Our HMM analysis of electroencephalogram signals uncovers interesting findings regarding differences in brain activity during sleep between those with and without SDB. These findings include stability of the percent time spent in HMM latent states across matched diseased and non-diseased groups and differences in the rate of transitioning.

Sleep-wake behavior in the rat: ultradian rhythms in a light-dark cycle and continuous bright light

Ultradian rhythms are a prominent but little-studied feature of mammalian sleep-wake and rest-activity patterns. They are especially evident in long-term records of behavioral state in polyphasic animals such as rodents. However, few attempts have been made to incorporate ultradian rhythmicity into models of sleep-wake dynamics, and little is known about the physiological mechanisms that give rise to ultradian rhythms in sleep-wake state. This study investigated ultradian dynamics in sleep and wakefulness in rats entrained to a 12-h:12-h light-dark cycle (LD) and in rats whose circadian rhythms were suppressed and free-running following long-term exposure to uninterrupted bright light (LL). We recorded sleep-wake state continuously for 7 to 12 consecutive days and used time-series analysis to quantify the dynamics of net cumulative time in each state (wakefulness [WAKE], rapid eye movement sleep [REM], and non-REM sleep [NREM]) in each animal individually. Form estimates and autocorrelation confirmed the presence of significant ultradian and circadian rhythms; maximum entropy spectral analysis allowed high-resolution evaluation of multiple periods within the signal, and wave-by-wave analysis enabled a statistical evaluation of the instantaneous period, peak-trough range, and phase of each ultradian wave in the time series. Significant ultradian periodicities were present in all 3 states in all animals. In LD, ultradian range was approximately 28% of circadian range. In LL, ultradian range was slightly reduced relative to LD, and circadian range was strongly attenuated. Ultradian rhythms were found to be quasiperiodic in both LD and LL. That is, ultradian period varied randomly around a mean of approximately 4 h, with no relationship between ultradian period and time of day.

Mixed effect Poisson log-linear models for clinical and epidemiological sleep hypnogram data

Bayesian Poisson log-linear multilevel models scalable to epidemiological studies are proposed to investigate population variability in sleep state transition rates. Hierarchical random effects are used to account for pairings of subjects and repeated measures within those subjects, as comparing diseased with non-diseased subjects while minimizing bias is of importance. Essentially, non-parametric piecewise constant hazards are estimated and smoothed, allowing for time-varying covariates and segment of the night comparisons. The Bayesian Poisson regression is justified through a re-derivation of a classical algebraic likelihood equivalence of Poisson regression with a log(time) offset and survival regression assuming exponentially distributed survival times. Such re-derivation allows synthesis of two methods currently used to analyze sleep transition phenomena: stratified multi-state proportional hazards models and log-linear generalized estimating equations (GEE) models for transition counts. An example data set from the Sleep Heart Health Study is analyzed. Supplementary material includes the analyzed data set as well as the code for a reproducible analysis.

Quantification of the fragmentation of rest-activity patterns in elderly individuals using a state transition analysis

OBJECTIVES

Recent interest in the temporal dynamics of behavioral states has spurred the development of analytical approaches for their quantification. Several analytical approaches for polysomnographic data have been described. However, polysomnography is cumbersome, perturbs behavior, and is limited to short recordings. Although less physiologically comprehensive than polysomnography, actigraphy is nonintrusive, amenable to long recordings, and suited to use in subjects' natural environments, and provides an indirect measure of behavioral state. We developed a probabilistic state transition model to quantify the fragmentation of human rest-activity patterns from actigraphic data. We then applied this to the study of the temporal dynamics of rest-activity patterns in older individuals.

DESIGN

Cross-sectional.

SETTING

Community-based.

PARTICIPANTS

621 community-dwelling individuals without dementia participating in the Rush Memory and Aging Project.

MEASUREMENTS AND RESULTS

We analyzed actigraphic data collected for up to 11 days. We processed each record to give a series of transitions between the states of rest and activity, calculated the probabilities of such transitions, and described their evolution as a function of time. From these analyses, we derived metrics of the fragmentation of rest or activity at scales of seconds to minutes. Regression modeling of the relationship of these metrics with clinical variables revealed significant associations with age, even after adjusting for sex, body mass index, and a broad range of medical comorbidities.

CONCLUSIONS

Probabilistic analyses of the transition dynamics of rest-activity data provide a high-throughput, automated, quantitative, and noninvasive method of assessing the fragmentation of behavioral states suitable for large scale human and animal studies; these methods reveal age-associated changes in the fragmentation of rest-activity patterns akin to those described using polysomnographic methods.

NREM sleep stage transitions control ultradian REM sleep rhythm

STUDY OBJECTIVES

The cyclic sequence of NREM and REM sleep, the so-called ultradian rhythm, is a highly characteristic feature of sleep. However, the mechanisms responsible for the ultradian REM sleep rhythm, particularly in humans, have not to date been fully elucidated. We hypothesize that a stage transition mechanism is involved in the determination of the ultradian REM sleep rhythm.

PARTICIPANTS

Ten healthy young male volunteers (AGE: 22 ± 4 years, range 19-31 years) spent 3 nights in a sleep laboratory. The first was the adaptation night, and the second was the baseline night. On the third night, the subjects received risperidone (1 mg tablet), a central serotonergic and dopaminergic antagonist, 30 min before the polysomnography recording.

MEASUREMENTS AND RESULTS

We measured and investigated transition probabilities between waking, REM, and NREM sleep stages (N1, N2, and N3) within the REM-onset intervals, defined as the intervals between the onset of one REM period and the beginning of the next, altered by risperidone. We also calculated the transition intensity (i.e., instantaneous transition rate) and examined the temporal pattern of transitions within the altered REM-onset intervals. We found that when the REM-onset interval was prolonged by risperidone, the probability of transitions from N2 to N3 was significantly increased within the same prolonged interval, with a significant delay and/or recurrences of the peak intensity of transitions from N2 to N3.

CONCLUSIONS

These results suggest that the mechanism governing NREM sleep stage transitions (from light to deep sleep) plays an important role in determining ultradian REM sleep rhythms.

Probabilistic sleep architecture models in patients with and without sleep apnea

Sleep fragmentation of any cause is disruptive to the rejuvenating value of sleep. However, methods to quantify sleep architecture remain limited. We have previously shown that human sleep-wake stage distributions exhibit multi-exponential dynamics, which are fragmented by obstructive sleep apnea (OSA), suggesting that Markov models may be a useful method to quantify architecture in health and disease. Sleep stage data were obtained from two subsets of the Sleep Heart Health Study database: control subjects with no medications, no OSA, no medical co-morbidities and no sleepiness (n = 374); and subjects with severe OSA (n = 338). Sleep architecture was simplified into three stages: wake after sleep onset (WASO); non-rapid eye movement (NREM) sleep; and rapid eye movement (REM) sleep. The connectivity and transition rates among eight 'generator' states of a first-order continuous-time Markov model were inferred from the observed ('phenotypic') distributions: three exponentials each of NREM sleep and WASO; and two exponentials of REM sleep. Ultradian REM cycling was accomplished by imposing time-variation to REM state entry rates. Fragmentation in subjects with severe OSA involved faster transition probabilities as well as additional state transition paths within the model. The Markov models exhibit two important features of human sleep architecture: multi-exponential stage dynamics (accounting for observed bout distributions); and probabilistic transitions (an inherent source of variability). In addition, the model quantifies the fragmentation associated with severe OSA. Markov sleep models may prove important for quantifying sleep disruption to provide objective metrics to correlate with endpoints ranging from sleepiness to cardiovascular morbidity.

Sleep disturbances in highly stress reactive mice: modeling endophenotypes of major depression

Background

Neuronal mechanisms underlying affective disorders such as major depression (MD) are still poorly understood. By selectively breeding mice for high (HR), intermediate (IR), or low (LR) reactivity of the hypothalamic-pituitary-adrenocortical (HPA) axis, we recently established a new genetic animal model of extremes in stress reactivity (SR). Studies characterizing this SR mouse model on the behavioral, endocrine, and neurobiological levels revealed several similarities with key endophenotypes observed in MD patients. HR mice were shown to have changes in rhythmicity and sleep measures such as rapid eye movement sleep (REMS) and non-REM sleep (NREMS) as well as in slow wave activity, indicative of reduced sleep efficacy and increased REMS. In the present study we were interested in how far a detailed spectral analysis of several electroencephalogram (EEG) parameters, including relevant frequency bands, could reveal further alterations of sleep architecture in this animal model. Eight adult males of each of the three breeding lines were equipped with epidural EEG and intramuscular electromyogram (EMG) electrodes. After recovery, EEG and EMG recordings were performed for two days.

Results

Differences in the amount of REMS and wakefulness and in the number of transitions between vigilance states were found in HR mice, when compared with IR and LR animals. Increased frequencies of transitions from NREMS to REMS and from REMS to wakefulness in HR animals were robust across the light-dark cycle. Detailed statistical analyses of spectral EEG parameters showed that especially during NREMS the power of the theta (6-9 Hz), alpha (10-15 Hz) and eta (16-22.75 Hz) bands was significantly different between the three breeding lines. Well defined distributions of significant power differences could be assigned to different times during the light and the dark phase. Especially during NREMS, group differences were robust and could be continuously monitored across the light-dark cycle.

Conclusions

The HR mice, i.e. those animals that have a genetic predisposition to hyper-activating their HPA axis in response to stressors, showed disturbed patterns in sleep architecture, similar to what is known from depressed patients. Significant alterations in several frequency bands of the EEG, which also seem to at least partly mimic clinical observations, suggest the SR mouse lines as a promising animal model for basic research of mechanisms underlying sleep impairments in MD.

Utility of sleep stage transitions in assessing sleep continuity

STUDY OBJECTIVES

Sleep continuity is commonly assessed with polysomnographic measures such as sleep efficiency, sleep stage percentages, and the arousal index. The aim of this study was to examine whether the transition rate between different sleep stages could be used as an index of sleep continuity to predict self-reported sleep quality independent of other commonly used metrics.

DESIGN AND SETTING

Analysis of the Sleep Heart Health Study polysomnographic data.

PARTICIPANTS

A community cohort.

MEASUREMENTS AND RESULTS

Sleep recordings on 5,684 participants were deemed to be of sufficient quality to allow visual scoring of NREM and REM sleep. For each participant, we tabulated the frequency of transitions between wake, NREM sleep, and REM sleep. An overall transition rate was determined as the number of all transitions per hour sleep. Stage-specific transition rates between wake, NREM sleep, and REM sleep were also determined. A 5-point Likert scale was used to assess the subjective experience of restless and light sleep the morning after the sleep study. Multivariable regression models showed that a high overall sleep stage transition rate was associated with restless and light sleep independent of several covariates including total sleep time, percentages of sleep stages, wake time after sleep onset, and the arousal index. Compared to the lowest quartile of the overall transition rate (<7.76 events/h), the odds ratios for restless sleep were 1.27, 1.42, and 1.38, for the second (7.77-10.10 events/h), third (10.11-13.34 events/h), and fourth (≥13.35 events/h) quartiles, respectively. Analysis of stage-specific transition rates showed that transitions between wake and NREM sleep were also independently associated with restless and light sleep.

CONCLUSIONS

Assessing overall and stage-specific transition rates provides a complementary approach for assessing sleep continuity. Incorporating such measures, along with conventional metrics, could yield useful insights into the significance of sleep continuity for clinical outcomes.

Correction for covariate measurement error in nonparametric longitudinal regression

We introduce a correction for covariate measurement error in nonparametric regression applied to longitudinal binary data arising from a study on human sleep. The data have been surveyed to investigate the association of some hormonal levels and the probability of being asleep. The hormonal effect is modeled flexibly while we account for the error-prone measurement of its concentration in the blood and the longitudinal character of the data. We present a fully Bayesian treatment utilizing Markov chain Monte Carlo inference techniques, and also introduce block updating to improve sampling and computational performance in the binary case. Our model is partly inspired by the relevance vector machine with radial basis functions, where usually very few basis functions are automatically selected for fitting the data. In the proposed approach, we implement such data-driven complexity regulation by adopting the idea of Bayesian model averaging. Besides the general theory and the detailed sampling scheme, we also provide a simulation study for the Gaussian and the binary cases by comparing our method to the naive analysis ignoring measurement error. The results demonstrate a clear gain when using the proposed correction method, particularly for the Gaussian case with medium and large measurement error variances, even if the covariate model is misspecified.

Sleep stage dynamics differ between children with and without obstructive sleep apnea

STUDY OBJECTIVES

Analysis of sleep dynamics--distributions of contiguous sleep and sleep stage durations--reveal exponential distributions and potential clinical utility in adults. We sought to examine these polysomnographic variables for the first time in children, and in the context of childhood sleep disordered breathing (SDB).

DESIGN AND SETTING

Analysis of polysomnographic data available from the Washtenaw County Adenotonsillectomy Cohort.

PARTICIPANTS

Selected subjects were 48 children aged 5-12 years with SDB (pediatric apnea/hypopnea index > or = 1.5) who were scheduled for adenotonsillectomy and 20 control subjects of similar ages without SDB. Subjects were studied at enrollment and again one year later in almost all cases.

RESULTS

Durations of sleep and specific sleep stage bouts generally followed exponential distributions. At baseline, the number of sleep stage changes, proportion of total sleep time occupied by stage 1 sleep, proportion stage 2 sleep, mean stage 2 duration, and mean stage REM duration each distinguished subjects with and without SDB (P < 0.05), but only mean stage 2 duration did so independently after accounting for the other variables (P = 0.03). At one-year follow-up, changes in total sleep time, mean stage 2 duration, and mean stage REM duration distinguished SDB from control subjects, but again only changes in mean stage 2 duration did so independently (P = 0.01).

CONCLUSIONS

Durations of uninterrupted sleep and specific sleep stages appear to follow exponential distributions in children with or without SDB. Parameters that describe these distributions--particularly mean duration of stage 2 sleep periods--may provide useful additions to standard sleep stage analyses.

Markov processes for the prediction of aircraft noise effects on sleep

BACKGROUND

Aircraft noise disturbs sleep and impairs recuperation. Authorities plan to expand Frankfurt airport.

OBJECTIVE

To quantitatively assess the effects of a traffic curfew (11 PM to 5 AM) at Frankfurt Airport on sleep structure.

DESIGN

Experimental sleep study; polysomnography for 13 consecutive nights.

SETTING

Sleep laboratory. Subjects. 128 healthy subjects, mean age (SD) 38 (13) years, range 19 to 65, 59% female. Intervention. Exposure to aircraft noise via loudspeakers.

MEASUREMENTS

A 6-state Markov state transition sleep model was used to simulate 3 noise scenarios with first-order Monte Carlo simulations: 1) 2005 traffic at Frankfurt Airport, 2) as simulation 1 but flights between 11 PM and 5 AM cancelled, and 3) as simulation 2, with flights between 11 PM and 5 AM from simulation 1 rescheduled to periods before 11 PM and after 5 AM. Probabilities for transitions between sleep stages were estimated with autoregressive multinomial logistic regression.

RESULTS

Compared to a night without curfew, models indicate small improvements in sleep structure in nights with curfew, even if all traffic is rescheduled to periods before and after the curfew period. For those who go to bed before 10:30 PM or after 1 AM, this benefit is likely to be offset by the expected increase of air traffic during late evening and early morning hours. Limitations. Limited ecologic validity due to laboratory setting and subject sample.

CONCLUSIONS

According to the decision analysis, it is unlikely that the proposed curfew at Frankfurt Airport substantially benefits sleep structure. Extensions of the model could be used to evaluate or propose alternative air traffic regulation strategies for Frankfurt Airport.

A new strategy to analyze possible association structures between dynamic nocturnal hormone activities and sleep alterations in humans

The human sleep process shows dynamic alterations during the night. Methods are needed to examine whether and to what extent such alterations are affected by internal, possibly time-dependent, factors, such as endocrine activity. In an observational study, we examined simultaneously sleep EEG and nocturnal levels of renin, growth hormone (GH), and cortisol (between 2300 and 0700) in 47 healthy volunteers comprising 24 women (41.67 ± 2.93 yr of age) and 23 men (37.26 ± 2.85 yr of age). Hormone concentrations were measured every 20 min. Conventional sleep stage scoring at 30-s intervals was applied. Semiparametric multinomial logit models are used to study and quantify possible time-dependent hormone effects on sleep stage transition courses. Results show that increased cortisol levels decrease the probability of transition from rapid-eye-movement (REM) sleep to wakefulness (WAKE) and increase the probability of transition from REM to non-REM (NREM) sleep, irrespective of the time in the night. Via the model selection criterion Akaike's information criterion, it was found that all considered hormone effects on transition probabilities with the initial state WAKE change with time. Similarly, transition from slow-wave sleep (SWS) to light sleep (LS) is affected by a “hormone-time” interaction for cortisol and renin, but not GH. For example, there is a considerable increase in the probability of SWS-LS transition toward the end of the night, when cortisol concentrations are very high. In summary, alterations in human sleep possess dynamic forms and are partially influenced by the endocrine activity of certain hormones. Statistical methods, such as semiparametric multinomial and time-dependent logit regression, can offer ambitious ways to investigate and estimate the association intensities between the nonstationary sleep changes and the time-dependent endocrine activities.

Rhythmicity in mice selected for extremes in stress reactivity: behavioural, endocrine and sleep changes resembling endophenotypes of major depression

BACKGROUND

Dysregulation of the hypothalamic-pituitary-adrenal (HPA) axis, including hyper- or hypo-activity of the stress hormone system, plays a critical role in the pathophysiology of mood disorders such as major depression (MD). Further biological hallmarks of MD are disturbances in circadian rhythms and sleep architecture. Applying a translational approach, an animal model has recently been developed, focusing on the deviation in sensitivity to stressful encounters. This so-called 'stress reactivity' (SR) mouse model consists of three separate breeding lines selected for either high (HR), intermediate (IR), or low (LR) corticosterone increase in response to stressors.

METHODOLOGY/PRINCIPLE FINDINGS

In order to contribute to the validation of the SR mouse model, our study combined the analysis of behavioural and HPA axis rhythmicity with sleep-EEG recordings in the HR/IR/LR mouse lines. We found that hyper-responsiveness to stressors was associated with psychomotor alterations (increased locomotor activity and exploration towards the end of the resting period), resembling symptoms like restlessness, sleep continuity disturbances and early awakenings that are commonly observed in melancholic depression. Additionally, HR mice also showed neuroendocrine abnormalities similar to symptoms of MD patients such as reduced amplitude of the circadian glucocorticoid rhythm and elevated trough levels. The sleep-EEG analyses, furthermore, revealed changes in rapid eye movement (REM) and non-REM sleep as well as slow wave activity, indicative of reduced sleep efficacy and REM sleep disinhibition in HR mice.

CONCLUSION/SIGNIFICANCE

Thus, we could show that by selectively breeding mice for extremes in stress reactivity, clinically relevant endophenotypes of MD can be modelled. Given the importance of rhythmicity and sleep disturbances as biomarkers of MD, both animal and clinical studies on the interaction of behavioural, neuroendocrine and sleep parameters may reveal molecular pathways that ultimately lead to the discovery of new targets for antidepressant drugs tailored to match specific pathologies within MD.

Bayesian semi parametric multi-state models

Multi-state models provide a unified framework for the description of the evolution of discrete phenomena in continuous time. One particular example is Markov processes which can be characterised by a set of time-constant transition intensities between the states. In this paper, we will extend such parametric approaches to semiparametric models with flexible transition intensities based on Bayesian versions of penalised splines. The transition intensities will be modelled as smooth functions of time and can further be related to parametric as well as nonparametric covariate effects. Covariates with time-varying effects and frailty terms can be included in addition. Inference will be conducted either fully Bayesian (using Markov chain Monte Carlo simulation techniques) or empirically Bayesian (based on a mixed model representation). A counting process representation of semiparametric multi-state models provides the likelihood formula and also forms the basis for model validation via martingale residual processes. As an application, we will consider human sleep data with a discrete set of sleep states such as REM and non-REM phases. In this case, simple parametric approaches are inappropriate since the dynamics underlying human sleep are strongly varying throughout the night and individual-specific variation has to be accounted for using covariate information and frailty terms.

Dynamics of sleep stage transitions in healthy humans and patients with chronic fatigue syndrome

Physiological and/or pathological implications of the dynamics of sleep stage transitions have not, to date, been investigated. We report detailed duration and transition statistics between sleep stages in healthy subjects and in others with chronic fatigue syndrome (CFS); in addition, we also compare our data with previously published results for rats. Twenty-two healthy females and 22 female patients with CFS, characterized by complaints of unrefreshing sleep, underwent one night of polysomnographic recording. We find that duration of deep sleep (stages III and IV) follows a power-law probability distribution function; in contrast, stage II sleep durations follow a stretched exponential and stage I, and REM sleep durations follow an exponential function. These stage duration distributions show a gradually increasing departure from the exponential form with increasing depth of sleep toward a power-law type distribution for deep sleep, suggesting increasing complexity of regulation of deeper sleep stages. We also find a substantial number of REM to non-REM sleep transitions in humans, while this transition is reported to be virtually nonexistent in rats. The relative frequency of this REM to non-REM sleep transition is significantly lower in CFS patients than in controls, resulting in a significantly greater relative transition frequency of moving from both REM and stage I sleep to awake. Such an alteration in the transition pattern suggests that the normal continuation of sleep in light or REM sleep is disrupted in CFS. We conclude that dynamic transition analysis of sleep stages is useful for elucidating yet-to-be-determined human sleep regulation mechanisms with pathophysiological implications.

Neurochemical regulation of sleep

This review summarizes recent developments in the field of sleep regulation, particularly in the role of hormones, and of synthetic GABA(A) receptor agonists. Certain hormones play a specific role in sleep regulation. A reciprocal interaction of the neuropeptides growth hormone (GH)-releasing hormone (GHRH) and corticotropin-releasing hormone (CRH) plays a key role in sleep regulation. At least in males GHRH is a common stimulus of non-rapid-eye-movement sleep (NREMS) and GH and inhibits the hypothalamo-pituitary adrenocortical (HPA) hormones, whereas CRH exerts opposite effects. Furthermore CRH may enhance rapid-eye-movement sleep (REMS). Changes in the GHRH:CRH ratio in favor of CRH appear to contribute to sleep EEG and endocrine changes during depression and normal ageing. In women, however, CRH-like effects of GHRH were found. Besides CRH somatostatin impairs sleep, whereas ghrelin, galanin and neuropeptide Y promote sleep. Vasoactive intestinal polypeptide appears to be involved in the temporal organization of human sleep. Beside of peptides, steroids participate in sleep regulation. Cortisol appears to promote REMS. Various neuroactive steroids exert specific effects on sleep. The beneficial effect of estrogen replacement therapy in menopausal women suggests a role of estrogen in sleep regulation. The GABA(A) receptor or GABAergic neurons are involved in the action of many of these hormones. Recently synthetic GABA(A) agonists, particularly gaboxadol and the GABA reuptake inhibitor tiagabine were shown to differ distinctly in their action from allosteric modulators of the GABA(A) receptor like benzodiazepines as they promote slow-wave sleep, decrease wakefulness and do not affect REMS.

Growth hormone-releasing hormone and corticotropin-releasing hormone enhance non-rapid-eye-movement sleep after sleep deprivation

The neuropeptides growth hormone (GH)-releasing hormone (GHRH) and corticotropin-releasing hormone (CRH) regulate sleep and nocturnal hormone secretion in a reciprocal fashion, at least in males. GHRH promotes sleep and GH and inhibits hypothalamo-pituitary-adrenocortical (HPA) hormones. CRH exerts opposite effects. In women, a sexual dimorphism was found because GHRH impairs sleep and stimulates HPA hormones. Sleep deprivation (SD) is the most powerful stimulus for inducing sleep. Studies in rodents show a key role of GHRH in sleep promotion after SD. The effects of GHRH and CRH on sleep-endocrine activity during the recovery night after SD are unknown. We compared sleep EEG, GH, and cortisol secretion between nights before and after 40 h of SD in 48 normal women and men aged 19-67 yr. During the recovery night, GHRH, CRH, or placebo were injected repetitively. After placebo during the recovery night, non-rapid-eye-movement sleep (NREMS) and rapid-eye-movement sleep (REMS) increased and wakefulness decreased compared with the baseline night. After GHRH, the increase of NREMS and the decrease of wakefulness were more distinct than after placebo. Also, after CRH, NREMS increased higher than after placebo, and a positive correlation was found between age and the baseline-related increase of slow-wave sleep. REMS increased after placebo and after GHRH, but not after CRH. EEG spectral analysis showed increases in the lower frequencies and decreases in the higher frequencies during NREMS after each of the treatments. Cortisol and GH did not differ between baseline and recovery nights after placebo. After GHRH, GH increased and cortisol decreased. Cortisol increased after CRH. No sex differences were found in these changes. Our data suggest that GHRH and CRH augment NREMS promotion after SD. Marked differences appear to exist in peptidergic sleep regulation between spontaneous and recovery sleep.

Dynamic analysis of multivariate failure time data

We present an approach for analyzing internal dependencies in counting processes. This covers the case with repeated events on each of a number of individuals, and more generally, the situation where several processes are observed for each individual. We define dynamic covariates, i.e., covariates depending on the past of the processes. The statistical analysis is performed mainly by the nonparametric additive approach. This yields a method for analyzing multivariate survival data, which is an alternative to the frailty approach. We present cumulative regression plots, statistical tests, residual plots, and a hat matrix plot for studying outliers. A program in R and S-PLUS for analyzing survival data with the additive regression model is available on the web site http://www.med.uio.no/imb/stat/addreg. The program has been developed to fit the counting process framework.

Sleep and the hypothalamo-pituitary-adrenocortical system

The intention of this review is to summarize the current knowledge on the bidirectional interaction between sleep EEG and the secretion of corticotropin (ACTH) and cortisol. The administration of various hypothalamic-pituitary- adrenocortical (HPA) hormones and their antagonists exerts specific sleep-EEG changes in several species including humans. It is well documented that corticotropin releasing hormone (CRH) impairs sleep and enhances vigilance. In addition, it may promote REM sleep. Changes in the growth hormone-releasing hormone (GHRH):CRH ratio in favour of CRH appear to contribute to shallow sleep, elevated cortisol levels and blunted GH in depression and ageing. On the other hand, in women GHRH appears to exert CRH-like effects on sleep. Acute cortisol administration increases slow-wave sleep (SWS) and GH, probably due to feedback inhibition of CRH, and inhibits REM sleep. With the mixed glucocorticoid and progesterone receptor antagonist mifepriston sleep is disrupted. Subchronic administration of the glucocorticoid agonist methylprednisolone desinhibited REM sleep. A synergism of elevated CRH and cortisol activity may contribute to REM disinhibition during depression. Also ACTH and vasopressin modulate sleep specifically but their physiological role remains unclear. For example acute icv vasopressin enhances wakefulness in rats, whereas its long-term administration increases SWS in the elderly. In various studies the interaction of sleep EEG and HPA hormones has been investigated at the baseline, after manipulation of sleep-wake behaviour and after environmental changes. Most studies agree that the circadian pattern of cortisol is relatively independent from sleep and environmental influences. Some data suggest a major effect of light on cortisol secretion. Sleeping is widely associated with blunting and awakenings are linked with increases of HPA hormones.