Hypomanic Personality and Sleep
Dissertation zur Erlangung des akademischen Grades Dr. med.
an der Medizinischen Fakultät der Universität Leipzig
eingereicht von David Wozniak geboren am 29.07.1990 in Berlin angefertigt an der Universität Leipzig in der Klinik für Psychiatrie und Psychotherapie
Betreuer: Prof. Dr. med. Ulrich Hegerl
Prof. Dr. med. Georg Schomerus (Klinikdirektor)
Ko-Betreuer: Dr. rer. nat. Tilman Hensch
Beschluss über die Verleihung des Doktorgrades vom: 14.07.2020
1
Table of contents
I)I)I) Introduction ...... 333
II)II)II) Publication ...... 101010
Supplementary Materials ...... 22202000
III)III)III) Summary ...... 33373777
IV)IV)IV) References ...... 44424222
V)V)V) Author contributionscontributions...... 44454555
VI)VI)VI) Declaration of authorship ...... 464646
VII)VII)VII) Curriculum vitae ...... 474747
VIII)VIII)VIII) Publications ...... 484848
IX)IX)IX) Acknowledgements ...... 494949
2
I)I)I) IntroductionIntroduction::::
Bipolar Disorder
Bipolar Disorder (BD) is a chronic mental disorder that occurs around 3% of the world's population and leads to increasing rates of cognitive and functional disability 1 and mortality 2. It is classified as an affective disorder with fluctuations in mood and behavior resulting in notable distress and affecting patient’s well-being.
The ICD-10 defines two types of BD. BD includes episodes of depression and mania
(Bipolar Disorder Type I) or episodes of depression and hypomania (Bipolar Disorder Type
II), the latter being defined as moderate phases of mood elevation distinguished by duration and severity to full-blown mania. While depressive episodes include negative mood, lack of energy and loss of interest, mania is characterized by elevated mood, restless psycho-motoric activity and impulsivity 3.
Disturbed Sleep in BD
Alternations in sleep quality were not only identified during depressive or manic episodes.
Meta-analysis of BD showed impairments in sleep latency, sleep duration, wake after sleep onset (WASO) and sleep efficiency 4. Additionally, sleep-wake variables showed a higher night-to-night variability during euthymic states 5.
Several datasets suggest that sleep impairment is not only a symptom of BD but an important factor contributing to the development of manic and hypomanic behavior itself.
Experimental and longitudinal studies with patients, healthy subjects and animals showed that sleep deprivation lead to mood changes in the following days and that disrupted sleep caused more severe manic symptoms 6-12 . The other way round, cognitive behavioral therapy for interepisodic insomnia aiming for sleep consolidation and improved sleep
3
efficiency resulted in a lower manic and hypomanic relapse 13 .
Thus, sleep has an important role for BD in many aspects - as a symptom, a contributing part to the development of the disorder or therapeutical intervention.
However, not only differed sleep itself is important. Excessive daytime sleepiness also plays an important role in BD. Daytime sleepiness without sufficient sleeping time has been descriped as an indicator for relapse to manic and hypomanic episodes in BD 16 .
Arousal Model of Affective Disorders and Bipolar Disorder
Daytime sleepiness has been characterized as the behavioral expression of an unstable
CNS-arousal 17 .
Arousal adapts to environmental stimulation and demands and is an individual condition from stable high arousal during an activated state (such as mental or physical effort) to unstable low arousal during a lower activation (such as mental and physical rest or even sleep). An EEG-based algorithm (Vigilance Algorithm Leipzig, VIGALL) allows objective measurement of arousal regulation 18-20 .
The Arousal Regulation Model of Affective Disorders and Attention Deficit Hyperactivity
Disorder (ADHD) suggests that unstable arousal in BD (and ADHD) can partly cause manic and hypomanic behavior. The initial situation of mania is seen in an unstable arousal caused by e.g. genetic disposition and sleep deficits. This triggers an autoregulatory behavioral syndrome with hyperactivity, sensation seeking and distractibility in order to stabilize the arousal, which leads to further sleep deficits and therefore even more unstable vigilance regulation. A vicious circle starts, finally resulting in full-blown mania 21-24 .
If impaired sleep is seen both as a symptom and as a predisposing trait in BD, sleep improvement might be a field for early intervention not only during euthymic or manic 4
episodes in patients already diagnosed with BD, but in individuals at high-risk of developing this mental disorder. Due to limitations of studying the sleep of BD patients in euthymic state, it remains unclear whether sleep impairments are a predisposition for manic and hypomanic states or whether they are merely a consequence of the disease process or due to medication since BD patients are often prescribed multiple medications impacting both arousal and sleep.
Sleep in healthy Subjects with vulnerability for Bipolar Disorder
Studying healthy subjects, who are vulnerable to BD, can improve our understanding of sleep impairment as a predisposing factor or as a mere symptom of BD.
Genetic high-risk studies have shown that sleep in offspring of BD is also impaired by poorer sleep quality, irregularity of sleep-wake times and higher WASO 25,26 . The limitations of many genetic studies lie not only in small sample sizes but also in the exclusive focus on offspring of patients who have actually fallen ill with BD.
The psychometric high-risk approach via questionnaires may be a useful step in order to overcome these limitations through larger sample sizes and less selection through focussing on subjects with parents with BD.
Psychometric High-Risk Approaches
The interest in psychometric high-risk approaches has a long tradition 27 and remains high nowadays since such approaches offer promising contributions to the diagnostics of mental disorders. Adequate early intervention before the development of a full-blown mental disorder could help improve the course of patients’ diseases and longterm outcomes.
5
Historical psychometric high-risk approaches were mainly focusing on psychosis. A longitudinal study showed that psychometric high-risk individuals for psychosis developed higher psychosis rates in the following 10 years 28 . Relevant clinical symptoms were reported more often in psychometric high-risk individuals for psychosis as well 29 . Meta- analysis of prognostic accuracy for psychometric high-risk individuals for psychosis showed a promising prognostic performance 30 being encouraging for further psychometric high-risk research in other mental disorders as well.
There have been several attempts for psychometric high-risk approaches for affective disorders as well. Typus Melancholicus was introduced to define a combination of traits consisting of being orderly and scrupulous and devoted to duty and family members 31-35 .
This combination of traits was assumed to be a risk-factor for endogenous depression and a possible predictor for the development of depressive episodes 36-38 .
The term Typus Manicus 39 , for example, was used as a description for a construct of certain personality traits of overactive and euphoric scope. Similar concepts are Hyperthymic
Personality 40 or Hypomanic Personality 41 . The latter concepts has been proposed as a vulnerability for developing BD.
Hypomanic Personality as a Vulnerability for Bipolar Disorder
The Hypomanic Personality Scale (HPS) is a psychometric tool for identifying individuals at high risk for BD 41.
The HPS is a 48-item self-report questionnaire to detect hypomanic traits. Hypomanic
Personality is a dimensional concept and is described as cheerful, equipped with a high self-esteem and ambition, who feels outstanding from others, optimistic, tireless, 6
extraverted and gregarious. At the same time, hypomanic attitude and behavior can be experienced as irresponsible, irritable, overbearing and reckless. Furthermore, subjects scoring high on HPS show a tendency to hypersexuality and drug use 41 .
Similar phenomena can be found during manic and hypomanic episodes in BD.
The Diagnostic and Statistical Manual of Mental Disorders describes these episodes as a temporary mood during a period of time, while otherwise the features of the Hypomanic
Personality Scale are thought as outlasting through life and habitual trait.
Several studies showed that people scoring high on HPS are vulnerable for developing BD, which could be acknowledged in later studies. While some authors describe Hypomanic
Personality being associated with a higher risk for mainly manic episodes 42-44 , others show a higher risk for both depressive and manic episodes 41,45. Most compelling is that the HPS predicted bipolar disorders in a 13-year follow-up study in so far not BD diagnosed subjects 46 .
The HPS allows not only the identification of unusually high scorers at the end of the continuum, but also the assessment of the entire normal variance. Such dimensional assessments of psychopathology have recently attracted much interest in the context of the endophenotype approach 47,48 and the Research Domain Criteria Project (RDoC) of the
National Institute of Mental Health (NIMH) 49 . Certain phenomena such as sleep impairments can be described not only in categorcial assessments like an actual manic episode, but can occur in dimensional states as in euthymic BD or high-risk individuals as provided by the HPS as well.
7
Hypomanic Personality and Sleep
To date, only two small studies using student samples have assessed the association between sleep and the HPS using sleep diaries and actigraphy 50,51 .
Shorter sleep duration was shown in HPS high scorers 50,51 . Also, greater night-to-night variability was observed in sleep duration and sleep efficiency 50,51 .
A deeper understanding of objective and subjective sleep of Hypomanic Personality seems necessary. Sleep of HPS high scorers seems to fit the findings of disturbed sleep in manic, hypomanic and euthymic BD. As the HPS is a dimensional psychometric questionnaire, it seems interesting to not only focus on extreme groups such as HPS high-scorers and low- scorers, who are already described in existing literature through the upper and lower decile of sum-scores, but to look at sleep alternations throughout the whole given results.
As sleep impairment shows a relevant link to BD, it is important to find out if it is cause of the disease or a predisposing factor. This study analyzed the associations of objective and subjective sleep parameters and daytime sleepiness with individuals at psychometric high risk for BD, the hypomanic personality. The HPS seems to detect high-risk individuals as promising targets for early intervention in sleep improvement in order to improve dimensional symptoms and contributing trigger for manic and hypomanic episodes.
To date, the findings in objective and subjective sleep of Hypomanic Personaltiy are limited and further investigations are needed.
In a large cohort of healthy subjects, we expected associations between the HPS and shorter and more disturbed sleep, increased daytime sleepiness and more night-to-night sleep variability.
8
II)II)II) PPPublicationPublicationublication::::
Vulnerability to Bipolar Disorder IIIsIs LLLinkedLinked to Sleep and Sleepiness
Tilman Hensch 1,2 *a, PhD; David Wozniak 1a , MD; Janek Spada 3, PhD; Christian Sander 1,2,3 , PhD; Christine Ulke 1,2,3 , MD, PhD; Dirk Wittekind 1, MD; Joachim Thiery 2,4 , MD, PhD; Markus Löffler 2,5 , MD, PhD, Philippe Jawinski 2,3,6 *b, PhD; Ulrich Hegerl 2,3,7b , MD, PhD
1 Department of Psychiatry and Psychotherapy, University of Leipzig Medical Center, Leipzig, Germany 2 LIFE—Leipzig Research Center for Civilization Diseases, Universität Leipzig, Leipzig, Germany 3 Depression Research Centre, German Depression Foundation, Leipzig, Germany
4 Institute of Laboratory Medicine, Clinical Chemistry and Molecular Diagnostics, University of Leipzig Medical Center, Leipzig, Germany
5 Institute for Medical Informatics, Statistics and Epidemiology (IMISE), University of Leipzig, Leipzig, Germany
6 Department of Psychology, Humboldt-Universität zu Berlin, Berlin, Germany
7 Department of Psychiatry, Psychosomatic Medicine and Psychotherapy, Goethe-Universität Frankfurt, Frankfurt am Main, Germany
*Corresponding authors; a co–first authors; b co-senior authors
Conflict of interest: The authors have nothing to disclose
Bibliography: Translational Psychiatry volume 9, Article number: 294 (2019), https://doi.org/10.1038/s41398-019-0632-1
9
Hensch et al. Translational Psychiatry (2019) 9:294 https://doi.org/10.1038/s41398-019-0632-1 Translational Psychiatry
ARTICLE Open Access Vulnerability to bipolar disorder is linked to sleep and sleepiness Tilman Hensch 1,2, David Wozniak1, Janek Spada3,ChristianSander1,2,3,ChristineUlke 1,2,3, Dirk Alexander Wittekind1, Joachim Thiery2,4, Markus Löffler2,5, Philippe Jawinski 2,3,6 and Ulrich Hegerl2,3,7
Abstract Sleep impairments are a hallmark of acute bipolar disorder (BD) episodes and are present even in the euthymic state. Studying healthy subjects who are vulnerable to BD can improve our understanding of whether sleep impairment is a predisposing factor. Therefore, we investigated whether vulnerability to BD, dimensionally assessed by the hypomanic personality scale (HPS), is associated with sleep disturbances in healthy subjects. We analyzed participants from a population-based cohort who had completed the HPS and had either a 7-day actigraphy recording or a Pittsburgh sleep quality index (PSQI) assessment. In addition, subjects had to be free of confounding diseases or medications. This resulted in 771 subjects for actigraphy and 1766 for PSQI analyses. We found strong evidence that higher HPS scores are associated with greater intraindividual sleep variability, more disturbed sleep and more daytime sleepiness. In addition, factor analyses revealed that core hypomanic features were especially associated with self-reported sleep impairments. Results support the assumption of disturbed sleep as a possibly predisposing factor for BD and suggest sleep improvement as a potential early prevention target. 1234567890():,; 1234567890():,; 1234567890():,; 1234567890():,; Introduction Further, in a recent study13 daytime sleepiness predicted Sleep disturbances and increased daytime sleepiness (hypo)manic relapse. Daytime sleepiness has been char- occur in bipolar disorder (BD), including in the euthymic acterized as the behavioral expression of an unstable state1,2. Accumulating evidence suggests that impaired or central nervous system (CNS)-arousal;14 and using elec- reduced sleep and increased daytime sleepiness are not troencephalogram (EEG), unstable arousal has been – only symptoms of BD, but contribute to the disease pro- demonstrated for BD, especially during mania15 18. The cess itself and to (hypo)manic behavior in particular. arousal regulation model of affective disorders and Experimental and longitudinal studies of patients, healthy attention-deficit/hyperactivity disorder (ADHD)15,17 takes subjects, and animals suggest that sleep deprivation can into consideration this unstable arousal in BD (and – induce (hypo)mania3 9. Accordingly, cognitive behavioral ADHD) and suggests that hypoarousal can partly cause therapy for inter-episode insomnia has resulted in a lower (hypo)manic behavior. In an autoregulatory manner, the (hypo)manic relapse rate10, and stabilization of sleep and hyperactive and sensation-seeking behavior is seen as a sleep–wake rhythms is an element of BD treatments11,12. compensatory arousal-stabilizing behavior, which can in turn increase sleep deficits, thus initiating a vicious circle contributing to mania15,17,18. Correspondence: Tilman Hensch ([email protected])or If impaired sleep is in fact a predisposing trait for BD, Philippe Jawinski ([email protected]) sleep improvement might be a target for early interven- 1Department of Psychiatry and Psychotherapy, University of Leipzig Medical Center, Leipzig, Germany tions. However, present findings on impaired sleep in 2LIFE—Leipzig Research Center for Civilization Diseases, Universität Leipzig, euthymic BD leave it unclear to what extent these dis- Leipzig, Germany turbances are a predisposition, a consequence of the Full list of author information is available at the end of the article. These authors contributed equally: Tilman Hensch, David Wozniak disease process, or due to medication. For example, BD These authors jointly supervised to this work: Philippe Jawinski, Ulrich Hegerl
© The Author(s) 2019 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 linktotheCreativeCommons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license 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 license, visit http://creativecommons.org/licenses/by/4.0/. Hensch et al. Translational Psychiatry (2019) 9:294 Page 2 of 10
patients are often prescribed numerous medications19, interview (TEMPS-I42) and self-rating instrument many of which impact both arousal and sleep. Studying (TEMPS-A43,44) in which the hyperthymic and cyclothy- healthy subjects who are vulnerable to BD can improve mic temperaments measure comparable constructs to the our understanding of whether sleep impairment is a HPS45,46. As is the case with the HPS, the TEMPS-A predisposing factor. Genetic high-risk studies have mostly temperaments show long-term stability47, aggregate in shown that sleep in healthy offspring of BD patients is families27,48, have been associated with psychiatric risk – indeed impaired20 22. However, in addition to the small genes49 and have been used to discriminate and predict sample sizes in many of these studies, a further inherent bipolar symptomatology48,50. One difference between the limitation is the exclusive focus on a particular subgroup, TEMPS-A and the HPS is that the TEMPS-A assesses a namely offspring of relatives who have fallen ill. There is hyperthymic temperament separately from a cyclothymic only one longitudinal study, which was not based on temperament, whereas in the HPS both aspects are genetic high-risk offspring but on a healthy community summed together in one total score. However, it has sample, and which could nonetheless predict subsequent recently been suggested that the HPS should be separated development of BD during a ten year follow up by poor into different subscales which may have divergent psy- – sleep quality at baseline23. Moreover, this prediction of chopathological correlates51 53. In addition to the BD by poor sleep persisted when excluding subjects with a research on affective temperament scales such as HPS and family history of affective disorders. However, results of TEMPS-A, longitudinal studies have contributed further this study were based on only 20 subjects who developed a evidence that subthreshold hypomanic symptoms often – full-blown BD. Missing data and a low probability of precede BD39,54 58. incident BD in healthy subjects not genetically at-risk are To date, only two small studies using student samples inherent problems of such longitudinal cohort studies. have assessed the association between sleep and HPS. In Thus, more data are needed. The psychometric high-risk HPS high-scorers, greater intraindividual variability in approach24, which assesses BD risk in large populations sleep duration was observed via sleep diaries59 and acti- via questionnaires such as the hypomanic personality graphy60. The latter study60 also showed shorter sleep scale (HPS)25, is one useful method. The HPS allows not duration and greater intraindividual variability in sleep only the identification of unusually high scorers (high risk efficiency in HPS high-scorers. As is usually the case, subjects), but also the assessment of the total normal these two studies utilized the HPS total sum-score. variance. The association of the entire continuum of However, in two recent studies51,53 HPS subscales were bipolarity with sleep disturbances can then be examined. derived from factor analyses which differed in their Such dimensional concepts of psychopathology have associations with psychopathologically relevant recently attracted much interest in the context of psy- traits51,53,61. – chiatric genetics and the endophenotype approach26 28, and also the Research Domain Criteria Project (RDoC) of Objectives the National Institute of Mental Health (NIMH) pointed The current study analyses the associations of objec- out the strength of dimensional assessments29. tive and subjective sleep parameters and daytime slee- The HPS assesses features of hypomanic episodes as piness with a risk factor for BD, the HPS. In a large described within the diagnostic systems, but also includes cohort of healthy subjects, we expect associations correlated traits such as social dominance. As the scale between the HPS and shorter and more disturbed sleep, assesses a temperamental dimension, i.e., a personality increased daytime sleepiness and more night-to-night trait rather than an acute episode, subjects are instructed sleep-variability. In contrast to most prior research, we to respond how they feel in general. Several studies have not only associate sleep with the HPS total score, but demonstrated the scale’s reliability and validity. The HPS also with HPS subscales. To this end, we conducted the is stable over time25,30, has been associated with psy- first factor analysis on the German translation62 of the chiatric risk genes31,32, aggregates in families of afflicted HPS. We hypothesize that the associations of the HPS patients33 and has discriminated bipolar patients from with sleep variables will differ with respect to the HPS controls34,35. Cross-sectionally, HPS high scorers showed subscales. higher rates of (hypo)manic and depressed episodes and more psychosocial impairment and substance-use25,36,37. Materials and methods Longitudinally, the HPS predicted BD and hypomanic Participants symptoms in addition to related disorders such as sub- The study sample was drawn from the LIFE-Adult – stance abuse38 40. study63, a population-based cohort comprised of 10,000 Further evidence for the role of hypomanic traits in BD inhabitants of the city of Leipzig, Germany. Of the total have been presented by other authors such as Hagop S. sample, 3031 participants aged 60–82 years completed the Akiskal and Richard A. Depue41. Akiskal developed an HPS. Subjects had to be free of diseases or medications Hensch et al. Translational Psychiatry (2019) 9:294 Page 3 of 10
which could strongly impact sleep–wake behavior. Thus, Results participants with a history of stroke, multiple sclerosis, Factorial structure of the HPS Parkinson’s disease, epilepsy, skull fracture, cerebral In the Supplementum, factor analysis results are tumor, or meningitis were excluded, leaving 2788 sub- described in detail and compared to the available factor jects. Further excluded were individuals reporting current solutions reported in two younger non-German sam- use of CNS-affecting drugs (leaving 2373 subjects). Based ples51,68. Results revealed three factors (see Supplemen- on data from structured clinical interviews for DSM-IV tary Figs. S1–S3), which were well in line with prior Axis I disorders, we selected subjects without a lifetime studies (see Supplementary Table S1). We labeled the first history of substance dependence, psychotic or BDs, and factor hypomanic core, as it was comprised of items with who were free of current affective or anxiety disorders clearly hypomanic content. Items loading high on the (leaving 2087 subjects). In addition, participants were second factor (social vitality) describe high self-con- required to have available data from either the Pittsburgh fidence, social dominance, and leadership. The third fac- sleep quality index (PSQI) assessment or an actigraphy tor appears to reflect a characteristic that has previously recording for at least five nights. This resulted in a final been recognized as “ordinariness”, as it describes balanced sample of 771 subjects for actigraphy association analyses and controlled people who consider themselves as average 68 (372 female, Mage = 70.3 y), and 1766 subjects for PSQI persons . Thus, we retained the term ordinariness. association analyses (835 female, Mage = 69.6 y). Factor analyses of the HPS were conducted using all 2861 sub- Correlation analyses jects with complete HPS responses (1371 female; age Table 1 shows the Spearman correlations between the range: 60–82 years; Mage = 70.0). The study was approved sleep parameters and HPS total as well as the subscales’ by Leipzig University’s Ethics Committee (263-2009- factor scores hypomanic core, social vitality, and ordi- 14122009) and subjects gave written informed consent. nariness. In total, 46 out of 84 correlations reached the level of significance, with 40 of them remaining significant Objective and subjective sleep assessment after multiple-test correction (see Supplementary Table To obtain objective measurements of sleep, subjects S2). Accordingly, the quantile–quantile plot suggests that wore the SenseWear Pro 3 Armband actigraph (Body- the distribution of observed p values considerably differs Media; Pittsburgh, Pennsylvania) for an average of from a p value distribution under the null hypothesis 6.9 days (range: 5–7 days). Analyses of night-sleep para- (Fig. 1). meters were carried out as described in detail else- – where64 66. Subjective ratings of sleep and sleep quality Objective sleep data were obtained using the German version of the PSQI67 a HPS total was significantly associated with shorter sleep self-rating instrument to assess sleep quality during the duration, a greater number of awakenings, more time past 4 weeks. awake after sleep-onset (WASO) and lower sleep effi- ciency. At the subscale level, these sleep variables were Hypomanic personality scale (HPS) also significantly associated in the same direction with The HPS, a self-rating scale used to assess BD risk25, hypomanic core, with the exception of number of awa- was developed in undergraduates, which is reflected in kenings, which narrowly missed the significance level. some of the 48 items. As the current study administered Associations of social vitality were similar, but only the HPS to elderly subjects, four items from the German WASO and sleep efficiency reached significance. Ordi- translation62 were deleted for reasons of compliance (see nariness, in contrast, was not correlated with any of these Supplementary Methods). actigraphic sleep variables. HPS total was even more strongly associated with the Statistical analysis night-to-night variability of the sleep parameters than The factorial structure of the HPS was analyzed using with the means. HPS total, hypomanic core and social the function irt.fa of R package psych (version 1.7.8) as vitality were all associated with greater intraindividual described in Supplementary Methods. Remaining statis- variability in sleep-onset time, sleep duration, number of tical analyses were performed using SPSS 22 (IBM; awakenings, WASO and sleep efficiency. In contrast, Armonk, New York). Associations between the sleep Ordinariness was associated with lower intraindividual variables and the HPS were conducted for both the HPS variability in the number of awakenings, WASO, and total sum-score (HPS total) and factor scores for each sleep efficiency. subscale. We conducted partial Spearman correlations adjusting for sex and age. In order to confirm the results, Subjective sleep data (PSQI) we additionally compared the top and bottom decile HPS HPS total correlated significantly with more daytime groups using Kruskal–Wallis tests. sleepiness and lower sleep quality. Even stronger Hensch et al. Translational Psychiatry (2019) 9:294 Page 4 of 10
Table 1 Partial Spearman correlations between hypomanic personality and sleep–wake variables
HPS total sum-score HPS subscale HPS subscale social HPS subscale hypomanic core vitality ordinariness
rho p rho p rho p rho p
Actigraphy (n = 771) Means Sleep-onset latency 0.038 0.294 0.042 0.240 0.054 0.134 −0.035 0.329 Sleep-onset time −0.006 0.864 −0.035 0.329 0.025 0.497 0.061 0.089 Sleep-offset time −0.013 0.721 −0.041 0.252 0.007 0.851 0.056 0.118 Sleep duration −0.079 0.029* −0.072 0.046* −0.059 0.101 0.034 0.350 NWAK 0.078 0.030* 0.069 0.056 0.069 0.056 −0.011 0.757 WASO 0.103 0.004** 0.091 0.011* 0.080 0.027* −0.046 0.203 Sleep efficiency −0.106 0.003** −0.101 0.005** −0.086 0.017* 0.033 0.361 Night-to-night variability Sleep-onset latency 0.019 0.594 −0.005 0.893 0.076 0.036* −0.026 0.470 Sleep-onset time 0.122 7E−4** 0.102 0.004** 0.113 0.002** 0.009 0.813 Sleep-offset time 0.110 0.002** 0.119 9E−4** 0.069 0.057 −0.062 0.087 Sleep duration 0.098 0.006** 0.092 0.011* 0.097 0.007** −0.003 0.931 NWAK 0.145 6E−5** 0.109 0.002** 0.152 2E−5** −0.092 0.011** WASO 0.115 0.001** 0.116 0.001** 0.095 0.008** −0.109 0.002** Sleep efficiency 0.106 0.003** 0.095 0.008** 0.103 0.004** −0.092 0.011** PSQI (n = 1766) Sleep-onset latencya 0.013 0.576 0.078 0.001** −0.080 8E−4** −0.098 4E−5** Bedtimea,b 0.029 0.230 0.018 0.444 0.041 0.087 0.001 0.968 Get-up timea −0.022 0.348 −0.023 0.344 −0.020 0.413 −0.013 0.581 Sleep durationa −0.038 0.114 −0.095 7E−5** 0.027 0.261 0.081 6E−4** Sleep efficiencyc 0.002 0.925 −0.065 0.006** 0.071 0.003** 0.091 1E−4** Daytime sleepinessa 0.075 0.002** 0.088 2E−4** 0.052 0.028* −0.086 3E−4** PSQI scored 0.063 0.008** 0.158 3E−11** −0.063 0.008** −0.165 3E−12**
Night-to-night variability is operationalized by intraindividual standard deviation (ISD) across a single subject’s multiple nights. Note that hypomanic core, social vitality and ordinariness here refer to factor scores derived from factor analyses as described in the Methods of the Supplementum. Results were additionally confirmed by analyses with traditional sum scores (see Supplementary Table S4). Effects of sex and age were partialled out. NWAK number of awakenings, WASO wake after sleep-onset time *p < 0.05 **p < 0.01 aBased on the respective Pittsburgh sleep quality index (PSQI) item bTime subject goes to bed cQuotient of sleep duration and time in bed with the latter calculated from bedtime and get-up time dPSQI total score calculated according the manual from all PSQI components; higher PSQI scores mean worse sleep quality
associations with PSQI-variables were found for hypo- sleepiness which did not significantly differ in its positive manic core, which was associated with a longer latency of associations with HPS total and hypomanic core. sleep-onset, shorter sleep duration, lower sleep efficiency, In contrast to HPS total and hypomanic core, social more daytime sleepiness, and lower sleep quality. The vitality, and ordinariness were associated with better obtained correlation coefficients for hypomanic core were subjective sleep and accordingly, both scales’ correlation significantly higher than those of the HPS total (Supple- coefficients differed significantly from HPS total and mentary Table S3), with the exception of daytime hypomanic core (Table S3). Social vitality and Hensch et al. Translational Psychiatry (2019) 9:294 Page 5 of 10
groups. Results resembled those of the correlation ana- 12 lyses (Supplementary Table S5), including the finding that 11 the HPS groups differed more strongly concerning the intraindividual night-to-night variability of the sleep 10 parameters than concerning the means. Figure 2 shows boxplots of the intraindividual sleep variability of acti- 9 graphic variables stratified by HPS extreme groups (see
scale) 8 Supplementary Fig. S4 for boxplots of all sleep variables). 10 With exception of sleep-onset latency, intraindividual 7 variability for all variables was significantly higher in the top HPS decile group compared with the bottom decile 6 group. 5 Discussion 4 The current study analyzed whether increased vulner- 3 ability to BD, as assessed by the HPS, is linked to more disturbed sleep in healthy subjects. Correlation analyses observed p-value (-log 2 were conducted between objective as well as subjective sleep parameters and HPS subscales and total scale. 1 Extreme group comparisons were also carried out to confirm results. 012 The correlation analyses as well as the extreme group comparisons revealed that a higher HPS total score is expected p-value (-log scale) 10 associated with worse sleep, greater night-to-night sleep Fig. 1 Permutation-based quantile–quantile plot showing that variability and more daytime sleepiness. Thus, results the observed p values (blue circles) considerably differ from a confirm findings of impaired sleep in genetic high risk random distribution under the null hypothesis (solid diagonal studies21. line). For the set of 84 observed p values (21 sleep variables × 4 personality scores), one million sets of 84 expected p values were Variability matters and HPS subscales differ in their sleep derived after data permutation. During data permutation, original correlations within the domain of hypomanic personality variables associations and the domain of sleep variables were preserved while original A noteworthy finding is that the HPS was more strongly correlations between the two domains were removed through associated with the night-to-night variability than with the random shuffling. Each set of p values was sorted in descending order. mean sleep variables. In line with this, irregularity in The solid diagonal line represents the mean expected p values at rank sleep/wake behavior has been reported for euthymic and 1–84 plotted against themselves. The upper and lower bound of the 2,69 gray area represent the 5th and 95th percentile of expected p values manic patients and subjects with genetically or psy- plotted against the mean expected p values. The blue circles represent chometrically operationalized heightened BD- the observed p values plotted against the mean expected p values risk2,21,59,60,70,71. In addition, another study found that greater sleep variability during euthymic state was asso- ciated with increased mania and depression severity over ordinariness were associated with shorter sleep-onset 12 months5. Thus, our results are in accordance with the latency, higher sleep efficiency, and better sleep quality. In role of social/circadian rhythm dysregulation in BD59,72. addition, ordinariness also correlated with longer sleep Another finding is that the HPS subscales differed in duration and lower daytime sleepiness, while social their associations with perceived sleep (PSQI). First, the vitality resembled hypomanic core in terms of its asso- correlations of hypomanic core with sleep were sig- ciation with more daytime sleepiness. nificantly higher than the correlations of HPS total with sleep. Second, and in contrast to HPS total and hypo- Additional analyses for confirmation of correlation results manic core, ordinariness, and social vitality showed Further analyses were carried out to confirm that results associations in the opposite direction (i.e., they were were not dependent on the data-analytic method. First, all associated with better perceived sleep). For ordinariness correlation analyses were repeated with unweighted sum- this was expected given the content of the scale. However, scores for each HPS subscale instead of factor scores, the association of social vitality with better perceived sleep which resulted in comparable, albeit, as expected, some- is noteworthy, because the subscale is not only positively what weaker associations (Supplementary Table S4). correlated with hypomanic core but also with objective Second, analyses were repeated using HPS decile extreme sleep impairment. Such diametric associations of the Hensch et al. Translational Psychiatry (2019) 9:294 Page 6 of 10
p = 0.2928 2.5 p = 0.0163 p = 0.0005 p = 0.0004 p = 0.0002 p = 0.0007 p = 0.0065 2.0 15
2.0 0.4 3 2.0 3 1.5
0.3 1.5 10 1.5 2 2 0.2 1.0 1.0 1.0 ISD of NWAK ISD of WASO (h) ISD of WASO 5
1 (h) ISD of sleep duration ISD of sleep efficiency (%) 0.1 ISD of sleep−onset time (h) 0.5 ISD of sleep−offset time (h) 1
ISD of sleep−onset latency (h) 0.5 0.5
0.0 0.0 0 HPS+ HPS− HPS+ HPS− HPS+ HPS− HPS+ HPS− HPS+ HPS− HPS+ HPS− HPS+ HPS−
Fig. 2 Boxplots of intraindividual night-to-night variability of actigraphic sleep variables stratified by HPS extreme groups. Intraindividual night-to-night variability is operationalized by intraindividual standard deviation (ISD) across a single subject’s multiple nights. Boxplots are stratified by top and bottom decile hypomanic personality scale (HPS) groups (HPS+, N = 63 vs. HPS−, N = 61). Boxes represent the interquartile range of each distribution (data between the lower and upper quartile), with the horizontal line corresponding to the median. Whiskers extend to the furthest observation within 1.5 times the interquartile range from the lower and upper quartile. Dots represent single data points, jittered horizontally to avoid overplotting
positively correlated HPS facets are in line with the two Sleep duration and daytime sleepiness available studies51,53 in which factor-analytic derived HPS The finding that HPS total and hypomanic core were subscales showed opposing associations with psycho- associated with shorter sleep duration is in line with studies pathologically relevant traits. It might be that subjects on naturally short sleepers who showed more hypomanic, scoring high on hypomanic core suffer more from extraverted and impulsive traits75,76.Thefinding is also impaired sleep, as the first factor (corresponding to consistent with one small study on young HPS high-scorers hypomanic core in our analyses) has been found to be and a study on children at genetic risk for BD. In both positively related to neuroticism51, whereas subjects studies, shorter sleep duration was also reported60,77.These scoring high on social vitality might not feel as impaired, findings of shorter sleep contrast with the meta-analyses given a negative association of social vitality with neuro- reporting longer sleep in euthymic BD compared to normal ticism51. Such reversed associations with HPS subscales controls1,2. This might imply that long sleep duration is not could cancel each other out in predicting sleep when a preexisting factor but occurs later as part of the disease. aggregated in the HPS total sum-score (as we demon- In addition, longer sleep could be merely a consequence of strated in a post-hoc linear regression analysis in Sup- the disease, as patients might spend more time in bed13 as a plementary Results). Thus, the current study strongly consequence of sedating medications, unemployment or supports the suggestion, recently made by several authors, the instruction to get adequate sleep in order to avoid – of utilizing HPS subscales51 53,61. mania recurrence. Finally, the meta-analyses’ findings of In line with the findings of impaired objective sleep in longer sleep during euthymia when measured with acti- both, hypomanic core and social vitality, both subscales graphy should be considered with caution, as sleep diaries were associated with increased daytime sleepiness. How- did not show longer sleep but only longer time-in-bed2.Itis ever, this increased sleepiness seems only to help subjects known that actigraphs based on accelerometers tend to scoring high on social vitality (presumably low in neu- overestimate sleep duration in cases of insomnia sympto- roticism) to quickly fall asleep, as was seen in the negative matology78. This might have contributed to the meta- correlation of social vitality with sleep-onset latency, analyses’ findings of longer sleep (as assessed by actigraphy) whereas hypomanic core was associated with longer sleep- in euthymic BD compared to healthy controls. One can onset latency. Thus, hypomanic core more strongly speculate that in the current study, because the applied resembles long sleep-onset latencies in (euthymic) BD1,2 SenseWear Armband uses not only an accelerometer, but and ADHD73,74, which may, despite high sleepiness, occur also temperature and galvanic skin response sensors to due to being too hyper and reluctant to fall asleep, estimate sleep, any potential overestimation of sleep dura- rumination, or circadian alterations. tion might have been less pronounced. Hensch et al. Translational Psychiatry (2019) 9:294 Page 7 of 10
The current study found that greater daytime sleepiness variety of reasons (e.g., life events, traveling, leisure associates with HPS, which is in line with findings of activities, circadian dysregulation, drugs, sleep apnea, or higher sleepiness scores in euthymic BD patients2. It has periodic limb movement disorder) have been suggested to been hypothesized that this sleepiness in BD reflects a contribute to (hypo)mania12. Finally, we focused our trait-like arousal instability in some BD patients15,16, analyses on sleep variables because our goal was to which would explain the impaired sustained attention in facilitate comparisons of our results for each given sleep the euthymic state and unaffected relatives79. In addition, variable, such as the WASO, with those from the available it has been speculated that the questionable efficacy and literature, particularly the meta-analyses of euthymic mania-inducing properties of antidepressants in BD might BD1,2. partly be attributable to their arousal-reducing properties, Nonetheless, one might ask whether new insights would which could be problematic if habitual arousal is already arise when all variables, including socioeconomic status, low15,80,81. One might argue that increased sleepiness age, sex, and body mass index are included simultaneously should result in longer sleep duration. However, it has in multivariable analyses. To investigate the relative been shown that in BD, longer sleep time and daytime importance of objective and subjective sleep variables sleepiness are independent presentations of hypersom- while accounting for other covariates in predicting HPS nia13,82. Thus, the current findings agree with the scores, we carried out a series of regression analyses. In assumption that hypoarousal plays a clinically relevant order to avoid overfitting and inflated R2 estimates, we role in BD. pursued a tenfold cross-validation approach which The association of HPS with shorter sleep and greater involves evaluating the performance of a prediction model daytime sleepiness is in line with the arousal regulation by applying the model to new data not used in training it. model of affective disorders and ADHD, which posits that Results, which are outlined in the Supplemental in more arousal instability due to short sleep or other reasons can detail, showed clear incremental predictive value of the contribute to (hypo)mania and ADHD15,17,83. As EEG is sleep variables over the covariates in predicting HPS total, an excellent tool to assess arousal84, we have developed hypomanic core, and ordinariness (Supplementary Table and validated the publically available Vigilance- S6). The incremental value of the sleep variables was most – Algorithm-Leipzig (VIGALL)14,85 89, which allows the pronounced for hypomanic core, which is in line with assessment of brain arousal in resting-EEG recordings. hypomanic core being most strongly correlated with sleep Utilizing VIGALL, we demonstrated lower arousal during in the correlation analyses. In contrast, there was no (hypo)mania18, and in genetic studies we demonstrated incremental predictive value of sleep variables on HPS for the first time a link between arousal and ion chan- social vitality. This finding could be due to the fact that a nels90,91, which are of high relevance for BD92. ADHD large proportion of variance in HPS social vitality was shows symptom overlap with mania and high comorbidity already explained by the covariate “socioeconomic status” with BD16,83. Correspondingly, several studies have also (see Supplementary Fig. S5), which might be considered shown both impaired sleep and hypoarousal in ADHD, as further evidence that social vitality may be the healthier which may partly explain attention deficits, response to facet of the HPS51 stimulants and compensatory hyperactive behavior15,83,93. As in mania, all factors which destabilize arousal such as Strengths and limitations sleep deficits worsen ADHD symptomatology, whereas The HPS has mostly been utilized in samples of ado- sleep improvement is of therapeutic value15,17,83. lescents and young adults. To increase the appropriate- ness of the HPS in our elderly subjects, we deleted four Prediction of HPS by sleep and covariates items, thus limiting the comparability of our findings with For several reasons, our analyses focused only on sleep other studies. However, the factorial structure of the variables. Thus, we did not include variables on the shortened HPS in our sample of German elderly adults diverse causes of disturbed sleep such as sleep apnea. One largely overlapped with those derived from the total HPS practical reason for this was that the LIFE cohort com- scale in younger samples51,53,68, suggesting that the uti- prises only limited data on the causes of disturbed sleep. lized version reveals comparable construct estimates even In fact, only two PSQI items provided information con- across different national and age cohorts. Nonetheless, cerning sleep apnea and periodic limb movement disorder our HPS subscales and their differential associations with (with these two items not aggregated into the PSQI total sleep variables need replication in other, particularly score). Thus, systematically accounting for sleep disorders younger samples. Concerning our subscales’ differential can only be done on a relatively weak basis. Most of all, we association with BD, we are currently lacking data, but focused on sleep variables because disturbed sleep has would hypothesize hypomanic core to be most strongly been suggested as the final common pathway for a variety associated with BD. There is a large overlap of items from of triggers for manic episodes9. Sleep disturbances for a our hypomanic core subscale with the items showing the Hensch et al. Translational Psychiatry (2019) 9:294 Page 8 of 10
largest correlations with BD diagnoses in a study by Miller sources of error variance, the highly consistent results are et al.34. All items from an HPS short version, comprising very compelling. the six items with the highest BD-association34, are part of our hypomanic core subscale. Conclusions The current study demonstrated that the HPS is also The associations of the HPS with worse sleep were applicable in subjects older than 60 years, which is not consistent, irrespective of the assessment modality and trivial given the strong decline of HPS scores with data analysis method. Sleep inter-night variability showed increasing age. For example, in our PSQI sample the mean the most pronounced associations with HPS, thus further of the HPS total score was M = 9.10 (SD = 5.37; mean qualifying as a characteristic variable in BD21,70,94,95.Asa extrapolated to a scale length of 48 items). The means and further practical implication, this study supports the – variances of younger samples are substantially higher. In the suggestion recently made by other authors51 53 that using undergraduate sample by Eckblad and Chapman25,the HPS subscales will increase the diagnostic power. mean was M = 21.74 (SD = 8.16). Similarly, in a German The association of sleep alterations with HPS supports sample62 with an age range of 17–30 years, the mean was assumptions derived from prior studies that sleep dis- M = 18 (SD = 8.91). Thus, the current results are compel- turbances may be a predisposing factor for BD. Analogous ling in light of the low HPS variance in our sample, which to prior studies in which sleep difficulties predicted mood mayhavemadeitmoredifficult to detect associations. episodes in genetic high risk off-spring70,77,96, one can From another perspective, the older age of our sample hypothesize that sleep might predict conversion to BD in may be both a limitation and a strength. The aim of our psychometric high-risk groups as assessed by the HPS. study was to associate the vulnerability factor HPS with Early interventions to prevent psychiatric disorders or sleep, and we were not interested in associations due to change their course are a cutting-edge topic in psy- effects of current or former affective episodes or treat- chiatry57. Thus, improving sleep in HPS high-scorers may ments. Therefore, we excluded subjects with acute affec- be a valuable early prevention approach, with sleep tive episodes, psychotropic drugs, or a life time diagnosis symptoms being easy to assess, modifiable, and largely of BD. In addition, our subjects are, due to their age, free of social stigmatization. relatively unlikely to develop BD in the future. This means Acknowledgements that, in contrast to studies in younger samples, we can This publication is supported by LIFE—Leipzig Research Center for Civilization also be quite sure that prodromes of emerging episodes Diseases, Universität Leipzig. LIFE is funded by means of the European Union, have not biased our results. However, the age and by the European Regional Development Fund (ERDF) and by means of the Free State of Saxony within the framework of the excellence initiative. This exclusion criteria of the current sample imply that we may publication was written within the framework of the cooperation between the have analyzed a sample with superior health, not com- German Depression Foundation and the “Deutsche Bahn Stiftung gGmbH”.We parable to the younger samples in which the HPS has thank Dr. Elise Paul for proofreading the paper. mostly been used. Notwithstanding, from a dimensional Author details perspective on psychopathology, it is still reasonable to 1Department of Psychiatry and Psychotherapy, University of Leipzig Medical associate the entire “remaining” vulnerability factor var- Center, Leipzig, Germany. 2LIFE—Leipzig Research Center for Civilization 3 iance with sleep. In addition, such a sample with superior Diseases, Universität Leipzig, Leipzig, Germany. Depression Research Centre, German Depression Foundation, Frankfurt am Main, Germany. 4Institute of health will still include subjects at high risk for BD. These Laboratory Medicine, Clinical Chemistry and Molecular Diagnostics, University subjects remained healthy, likely because of a lack of of Leipzig Medical Center, Leipzig, Germany. 5Institute for Medical Informatics, triggering factors or life-events or because of additional Statistics and Epidemiology (IMISE), University of Leipzig, Leipzig, Germany. 6Department of Psychology, Humboldt-Universität zu Berlin, Berlin, Germany. protective traits. It is remarkable that in this sample of 7Department of Psychiatry, Psychosomatic Medicine and Psychotherapy, likely superior health, the HPS is still associated with Goethe-Universität Frankfurt, Frankfurt am Main, Germany more impaired sleep and greater sleepiness. Conflict of interest The present study was well-powered and revealed The authors declare that they have no conflict of interest. strong evidence for a link between hypomanic tempera- ment and sleep alterations. Nevertheless, observed effect sizes, albeit substantial in extreme group comparisons, Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in were small in correlation analyses across all subjects. One published maps and institutional affiliations. reason for this might be that elderly subjects show low variances in HPS and wake–sleep behavior. Further, Supplementary Information accompanies this paper at (https://doi.org/ during the 7-day actigraphy assessment, sleep was 10.1038/s41398-019-0632-1). undoubtedly influenced by numerous factors beyond the Received: 15 March 2019 Revised: 1 October 2019 Accepted: 20 October subject’s predisposition (e.g., common cold, sleep of the 2019 partner, and reactivity to the actigraph). Given these Hensch et al. Translational Psychiatry (2019) 9:294 Page 9 of 10
References 33. Meyer, T. D. & Hautzinger, M. Hypomanic personality, social anhedonia and 1. Geoffroy, P. A. et al. Sleep in patients with remitted bipolar disorders: a meta- impulsive nonconformity: evidence for familial aggregation? J. Personal. Disord. analysis of actigraphy studies. Acta Psychiatr. Scand. 131,89–99 (2015). 15,281–299 (2001). 2. Ng, T. H. et al. Sleep-wake disturbance in interepisode bipolar disorder and 34. Miller, C. J., Johnson, S. L., Kwapil, T. R. & Carver, C. S. Three studies on self- high-risk individuals: a systematic review and meta-analysis. Sleep. Med. Rev. 20, report scales to detect bipolar disorder. J. Affect. Disord. 128, 199–210 46–58 (2015). (2011). 3. Barbini, B., Bertelli, S., Colombo, C. & Smeraldi, E. Sleep loss, a possible factor in 35. Rowland, J. E. et al. Cognitive regulation of negative affect in schizophrenia augmenting manic episode. Psychiatry Res. 65,121–125 (1996). and bipolar disorder. Psychiatry Res. 208,21–28 (2013). 4. Bauer, M. et al. Temporal relation between sleep and mood in patients with 36. Klein, D. N., Lewinsohn, P. M. & Seeley, J. R. Hypomanic personality traits in a bipolar disorder. Bipolar Disord. 8,160–167 (2006). community sample of adolescents. J. Affect. Disord. 38,135–143 (1996). 5. Gruber, J. et al. Sleep matters: sleep functioning and course of illness in bipolar 37. Meyer, T. D. & Hautzinger, M. Screening for bipolar disorders using the disorder. J. Affect. Disord. 134,416–420 (2011). Hypomanic Personality Scale. J. Affect. Disord. 75,149–154 (2003). 6. Lewis, K. S. et al. Sleep loss as a trigger of mood episodes in bipolar disorder: 38. Kwapil, T. R. et al. A longitudinal study of high scorers on the hypomanic individual differences based on diagnostic subtype and gender. Br.J.Psy- personality scale. J. Abnorm. Psychol. 109, 222–226 (2000). chiatry 211,169–174 (2017). 39.Walsh,M.A.,DeGeorge,D.P.,Barrantes-Vidal,N.&Kwapil,T.R.A3-year 7. Sylvia, L. G. et al. Sleep disturbance in euthymic bipolar patients. J. Psycho- longitudinal study of risk for bipolar spectrum psychopathology. J. Abnorm. pharmacol. 26,1108–1112 (2012). Psychol. 124,486–497 (2015). 8. Wehr, T. A. Sleep loss: a preventable cause of mania and other excited states. J. 40. Blechert,J.&Meyer,T.D.Aremeasures of hypomanic personality, impulsive Clin. Psychiatry 50(Suppl), 8–16 (1989). discussion 45-17. nonconformity and rigidity predictors of bipolar symptoms? Br.J.Clin.Psychol. 9. Wehr,T.A.,Sack,D.A.&Rosenthal,N.E.Sleepreductionasafinal common 44,15–27 (2005). pathway in the genesis of mania. Am.J.Psychiatry144, 201–204 (1987). 41. Depue,R.A.,Krauss,S.,Spoont,M.R.&Arbisi,P.Generalbehaviorinventory 10. Harvey, A. G. et al. Treating insomnia improves mood state, sleep, and func- identification of unipolar and bipolar affective conditions in a nonclinical tioning in bipolar disorder: a pilot randomized controlled trial. J. Consult Clin. university population. J. Abnorm. Psychol. 98,117–126 (1989). Psychol. 83, 564–577 (2015). 42. Akiskal,H.S.etal.TEMPS-I:delineating the most discriminant traits of the 11. Barbini, B. et al. Dark therapy for mania: a pilot study. Bipolar Disord. 7,98–101 cyclothymic, depressive, hyperthymic and irritable temperaments in a non- (2005). patient population. J. Affect. Disord. 51,7–19 (1998). 12. Plante, D. T. & Winkelman, J. W. Sleep disturbance in bipolar disorder: ther- 43. Akiskal,H.S.&Akiskal,K.K.TEMPS:temperament evaluation of Memphis, Pisa, apeutic implications. Am.J.Psychiatry165, 830–843 (2008). Paris and San Diego. J. Affect. Disord. 85,1–2 (2005). 13. Kaplan, K. A. et al. Hypersomnia subtypes, sleep and relapse in bipolar disorder. 44. Akiskal,H.S.,Akiskal,K.K.,Haykal,R.F.,Manning,J.S.&Connor,P.D.TEMPS-A: Psychol. Med. 45, 1751–1763 (2015). progress towards validation of a self-rated clinical version of the temperament 14. Jawinski, P. et al. Recorded and reported sleepiness: the association between evaluation of the Memphis, Pisa, Paris, and San Diego Autoquestionnaire. J. brain arousal in resting state and subjective daytime sleepiness. Sleep 40, Affect. Disord. 85,3–16 (2005). zsx099, https://doi.org/10.1093/sleep/zsx099 (2017). 45. Hensch, T., Herold, U. & Brocke, B. An electrophysiological endophenotype 15. Hegerl, U. & Hensch, T. The vigilance regulation model of affective disorders of hypomanic and hyperthymic personality. J. Affect. Disord. 101,13–26 and ADHD. Neurosci. Biobehav. Rev. 44,45–57 (2014). (2007). 16. Hegerl, U., Himmerich, H., Engmann, B. & Hensch, T. Mania and attention- 46. Walsh,M.A.,Royal,A.M.,Barrantes-Vidal,N.&Kwapil,T.R.Theassociationof deficit/hyperactivity disorder: common symptomatology, common patho- affective temperaments with impairment and psychopathology in a young physiology and common treatment? Curr. Opin. Psychiatry 23,1–7(2010). adult sample. J. Affect. Disord. 141,373–381 (2012). 17. Hegerl, U., Sander C. & Hensch T. Arousal regulation in affective disorders. In: 47. Kawamura, Y. et al. Six-year stability of affective temperaments as measured by Systems Neuroscience in Depression (ed Frodl T.). (Academic Press, 2016). TEMPS-A. Psychopathology 43,240–247 (2010). 18. Wittekind,D.A.etal.Earlyreportonbrainarousalregulationinmanicvs 48. Solmi, M. et al. A comparative meta-analysis of TEMPS scores across mood depressive episodes in bipolar disorder. Bipolar Disord. 18,502–510 (2016). disorder patients, their first-degree relatives, healthy controls, and other psy- 19. Ghaemi, S. N. et al. Pharmacological treatment patterns at study entry for the chiatric disorders. J. Affect. Disord. 196,32–46 (2016). first 500 STEP-BD participants. Psychiatr. Serv. 57, 660–665 (2006). 49. Greenwood,T.A.,Akiskal,H.S.,Akiskal,K.K.&Kelsoe,J.R.Genome-wide 20. Maoz, H. et al. Dimensional psychopathology in preschool offspring of parents association study of temperament in bipolar disorder reveals significant with bipolar disorder. J. Child Psychol. Psychiatry 55,144–153 (2014). associations with three novel Loci. Biol. Psychiatry 72,303–310 (2012). 21. Melo, M. C. et al. Sleep and circadian alterations in people at risk for bipolar 50. DeGeorge, D. P., Walsh, M. A., Barrantes-Vidal, N. & Kwapil, T. R. A three-year disorder: a systematic review. J. Psychiatr. Res. 83,211–219 (2016). longitudinal study of affective temperaments and risk for psychopathology. J. 22. Pancheri, C. et al. A systematic review on sleep alterations anticipating the Affect. Disord. 164,94–100 (2014). onset of bipolar disorder. Eur. Psychiatry 58,45–53 (2019). 51. Schalet,B.D.,Durbin,C.E.&Revelle,W.Multidimensionalstructureofthe 23. Ritter, P. S. et al. Disturbed sleep as risk factor for the subsequent onset of Hypomanic Personality Scale. Psychol. Assess. 23,504–522 (2011). bipolar disorder-Data from a 10-year prospective-longitudinal study among 52. Stanton,K.,Gruber,J.&Watson,D.Basicdimensionsdefining mania risk: a adolescents and young adults. J. Psychiatr. Res. 68,76–82 (2015). structural approach. Psychol. Assess. 29, 304–319 (2017). 24. Miller, G. A. The Behavioral High-Risk Paradigm in Psychopathology.(Springer, 53. Terrien, S., Stefaniak, N., Morvan, Y. & Besche-Richard, C. Factor structure of the 1995). French version of the Hypomanic Personality Scale (HPS) in non-clinical young 25. Eckblad, M. & Chapman, L. J. Development and validation of a scale for adults. Compr. Psychiatry 62,105–113 (2015). hypomanic personality. J. Abnorm. Psychol. 95,214–222 (1986). 54. Alloy, L. B. et al. Progression along the bipolar spectrum: a longitudinal study 26. Burmeister,M.,McInnis,M.G.&Zollner,S.Psychiatricgenetics:progressamid of predictors of conversion from bipolar spectrum conditions to bipolar I and controversy. Nat. Rev. Genet. 9,527–540 (2008). II disorders. J. Abnorm. Psychol. 121,16–27 (2012). 27. Savitz, J. B. & Ramesar, R. S. Personality: is it a viable endophenotype for genetic 55. Tijssen, M. J. et al. Evidence that bipolar disorder is the poor outcome fraction studies of bipolar affective disorder? Bipolar Disord. 8,322–337 (2006). of a common developmental phenotype: an 8-year cohort study in young 28. Smoller, J. W. et al. Psychiatric genetics and the structure of psychopathology. people. Psychol. Med. 40, 289–299 (2010). Mol. Psychiatry 24,409–420 (2019). 56. van Os, J. The dynamics of subthreshold psychopathology: implications for 29. Cuthbert, B. N. Research Domain Criteria: toward future psychiatric nosologies. diagnosis and treatment. Am. J. Psychiatry 170,695–698 (2013). Dialogues Clin. Neurosci. 17,89–97 (2015). 57. Vieta, E. et al. Early intervention in bipolar disorder. Am.J.Psychiatry175, 30. Hofmann,B.U.&Meyer,T.D.Moodfluctuations in people putatively at risk for 411–426 (2018). bipolar disorders. Br. J. Clin. Psychol. 45, 105–110 (2006). 58. Hafeman, D. M. et al. Assessment of a person-level risk calculator to predict 31. Johnson, S. L., Carver, C. S., Joormann, J. & Cuccaro, M. A genetic analysis of the new-onset bipolar spectrum disorder in youth at familial risk. JAMA Psychiatry validity of the Hypomanic Personality Scale. Bipolar Disord. 17,331–339 (2015). 74,841–847 (2017). 32. Ortega-Alonso, A. et al. Genome-wide association study of psychosis prone- 59. Meyer, T. D. & Maier, S. Is there evidence for social rhythm instability in people ness in the Finnish population. Schizophrenia Bull. 43, 1304–1314 (2017). at risk for affective disorders? Psychiatry Res. 141,103–114 (2006). Hensch et al. Translational Psychiatry (2019) 9:294 Page 10 of 10
60. Ankers, D. & Jones, S. H. Objective assessment of circadian activity and sleep 77. Egeland,J.A.etal.A16-yearprospective study of prodromal features prior to patterns in individuals at behavioural risk of hypomania. J. Clin. Psychol. 65, BPI onset in well Amish children. J. Affect. Disord. 142,186–192 (2012). 1071–1086 (2009). 78. Lichstein, K. L. et al. Actigraphy validation with insomnia. Sleep 29,232–239 61. Ford,B.Q.,Mauss,I.B.&Gruber,J.Valuinghappinessisassociatedwithbipolar (2006). disorder. Emotion 15,211–222 (2015). 79. Bora, E., Yucel, M. & Pantelis, C. Cognitive endophenotypes of bipolar disorder: 62. Meyer, T. D., Drüke, B. & Hautzinger, M. Hypomane Persönlichkeit-Psy- a meta-analysis of neuropsychological deficits in euthymic patients and their chometrische Evaluation und erste Ergebnisse zur Validität der deut- first-degree relatives. J. Affect. Disord. 113,1–20 (2009). schen Version der Chapman-Skala. Z. Klin. Psychol. Psychother. 29,35–42 80. Hensch, T. et al. Yawning in depression: worth looking into. Pharma- (2000). copsychiatry 48, 118–120 (2015). 63. Loeffler, M. et al. The LIFE-Adult-Study: objectives and design of a population- 81. Ulke, C. et al. Brain arousal regulation in SSRI-medicated patients with major based cohort study with 10,000 deeply phenotyped adults in Germany. BMC depression. J. Psychiatr. Res. 108,34–39 (2018). Public Health 15, 691 (2015). 82. Kaplan, K. A., Gruber, J., Eidelman, P., Talbot, L. S. & Harvey, A. G. Hypersomnia in 64.Spada,J.etal.Geneticassociationof objective sleep phenotypes with a inter-episode bipolar disorder: does it have prognostic significance? J. Affect. functional polymorphism in the neuropeptide S receptor gene. PLoS ONE 9, Disord. 132,438–444 (2011). e98789 (2014). 83. Geissler, J., Romanos, M., Hegerl, U. & Hensch, T. Hyperactivity and sensation 65. Spada, J. et al. Genome-wide association analysis of actigraphic sleep phe- seeking as autoregulatory attempts to stabilize brain arousal in ADHD and notypes in the LIFE Adult Study. J. Sleep. Res. 25,690–701 (2016). mania? Atten. Defic. Hyperact. Disord. 6,159–173 (2014). 66. Jawinski, P. et al. Time to wake up: no impact of COMT Val158Met gene 84. Sander,C.,Hensch,T.,Wittekind,D.A.,Bottger,D.&Hegerl,U.Assessmentof variation on circadian preferences, arousal regulation and sleep. Chronobiol. Int. wakefulnessandbrainarousalregulation in psychiatric research. Neu- 33,893–905 (2016). ropsychobiology 72,195–205 (2015). 67. Hinz, A. et al. Sleep quality in the general population: psychometric properties 85. Huang, J. et al. Evoked potentials and behavioral performance during different of the Pittsburgh Sleep Quality Index, derived from a German community states of brain arousal. BMC Neurosci. 18,21,https://doi.org/10.1186/s12868- sample of 9284 people. Sleep Med. 30,57–63 (2017). 017-0340-9 (2017). 68. Rawlings, D., Barrantes-Vidal, N., Claridge, G., McCreery, C. & Galanos, G. A factor 86. Huang, J. et al. Test-retest reliability of brain arousal regulation as assessed with analytic study of the Hypomanic Personality Scale in British, Spanish and VIGALL 2.0. Neuropsychiatr. Electrophysiol. 1,1–13 (2015). Australian samples. Personal. Individ. Differ. 28,73–84 (2000). 87. Olbrich, S. et al. EEG-vigilance and BOLD effect during simultaneous EEG/fMRI 69. Scott, J. et al. Activation in bipolar disorders: a systematic review. JAMA Psy- measurement. Neuroimage 45,319–332 (2009). chiatry 74,189–196 (2017). 88. Huang, J. et al. Impact of brain arousal and time-on-task on autonomic ner- 70. Duffy, A., Jones S., Goodday S. & Bentall R. Candidate risks indicators for bipolar vous system activity in the wake-sleep transition. BMC Neurosci. 19, https://doi. disorder: early intervention opportunities in high-risk youth. Int. J. Neu- org/10.1186/s12868-018-0419-y (2018). ropsychopharmacol. 19, pyv071, https://doi.org/10.1093/ijnp/pyv071 (2015). 89. Ulke, C. et al. Coupling and dynamics of cortical and autonomic signals are 71. Hoaki, N. et al. Biological aspect of hyperthymic temperament: light, sleep, and linked to central inhibition during the wake-sleep transition. Sci. Rep. 7, 11804 serotonin. Psychopharmacology 213, 633–638 (2011). (2017). 72. Alloy, L. B., Nusslock, R. & Boland, E. M. The development and course of bipolar 90. Jawinski, P. et al. Human brain arousal in the resting state: a genome-wide spectrum disorders: an integrated reward and circadian rhythm dysregulation association study. Mol. Psychiatry 24,1599–1609 (2019). model. Annu Rev. Clin. Psychol. 11, 213–250 (2015). 91. Jawinski, P. et al. Brain arousal regulation in carriers of bipolar disorder risk 73. Díaz-Román, A., Mitchell, R. & Cortese,S.SleepinadultswithADHD:systematic alleles. Neuropsychobiology 72,65–73 (2015). review and meta-analysis of subjective and objective studies. Neurosci. Bio- 92. Harrison, P. J., Geddes, J. R. & Tunbridge, E. M. The emerging neurobiology of behav. Rev. 89,61–71 (2018). bipolar disorder. Trends Neurosci. 41,18–30 (2018). 74. Cortese, S., Faraone, S. V., Konofal, E. & Lecendreux, M. Sleep in children 93. Strauss, M. et al. Brain arousal regulation in adults with attention-deficit/ with attention-deficit/hyperactivity disorder: meta-analysis of subjective hyperactivity disorder (ADHD). Psychiatry Res. 261,102–108 (2018). and objective studies. J. Am. Acad. Child Adolesc. Psychiatry 48, 894–908 94. Kaufmann, C. N., Gershon, A., Eyler, L. T. & Depp, C. A. Clinical significance of (2009). mobile health assessed sleep duration and variability in bipolar disorder. J. 75. Monk,T.H.,Buysse,D.J.,Welsh,D.K.,Kennedy,K.S.&Rose,L.R.Asleepdiary Psychiatr. Res. 81,152–159 (2016). and questionnaire study of naturally short sleepers. J. Sleep. Res. 10,173–179 95. Scott,J.,Vaaler,A.E.,Fasmer,O.B.,Morken,G.&Krane-Gartiser,K.Apilotstudy (2001). to determine whether combinations of objectively measured activity para- 76. Curtis, B. J., Williams P. G. & Anderson J. S. Objective cognitive functioning in meters can be used to differentiate between mixed states, mania, and bipolar self-reported habitual short sleepers not reporting daytime dysfunction: depression. Int J. Bipolar Disord. 5,5(2017). examination of impulsivity via delay discounting. Sleep 41,zsy115,https://doi. 96. Levenson, J. C. et al. Longitudinal sleep phenotypes among offspring of org/10.1093/sleep/zsy115 (2018). bipolar parents and community controls. J. Affect. Disord. 215,30–36 (2017).
Supplementary Material
Supplementary Methods Participants Hypomanic Personality ScScaleale (HPS) HPS factor analysis
Supplementary Results Factorial structure of the HPS Supplementary Fig. S1 Scree plot showing the eigenvalues of the principal factors derived from the tetrachoric correlation matrix of HPS items Supplementary Fig. S2 ICLUST output based on the tetrachoric correlation matrix of HPS items Supplementary Fig. S3 Tetrachoric correlation matrix of the HPS items Supplementary Table S1 Loadings and IRT parameters of the HPS three-factor model
Correlation analyses Supplementary Table S2 Partial Spearman correlations between HPS and sleep-wake variables. FDR- corrected and nominal p-values are given for comparison Supplementary Table S3 Comparison of Spearman correlations
Secondary analyses for confirmation of results Supplementary Table S4 Partial Spearman correlations between HPS (unweighted sum scores) and sleep- wake variables Supplementary Table S5 HPS total sum score and sleep-wake variables – extreme group comparisons Supplementary Fig. S4 Boxplots of actigraphy and PSQI variables stratified by HPS extreme groups
Exploratory regression analysis with HPS subscales
Prediction of HPS by sleep and covariates Supplementary Table S6 Pearson correlations between predicted and measured HPS scores Supplementary Fig. S5 Bar plot showing the relative importance of each variable to the prediction model
Supplementary References
20
Supplementary Methods
Hypomanic Personality Scale (HPS) We administered the German translation 1 of the Hypomanic Personality Scale (HPS) 2. The HPS is a self-rating instrument and is comprised of 48 dichotomous items. As the current study utilized the HPS in elderly subjects aged above 60 years for the first time, the following four inadequate items were deleted for reasons of tolerance and compliance: “I am frequently so ‘hyper’ that my friends kiddingly ask me what drug I’m taking”, “I expect that someday I will succeed in several different professions”, “There are so many fields I could succeed in that it seems a shame to have to pick”, “A hundred years after I’m dead, my achievements will probably have been forgotten”. Our HPS total sum score, with a possible maximum of 44, ranged from 0 to 37 (mean: 8.34, SD: 5.37) in the PSQI sample and from 0 to 37 (mean: 8.55, SD: 5.59) in the actigraphy sample. Extrapolated to a scale comprised of 48 items, the means were M = 9.10 and M = 9.33, respectively.
HPS factor analysis Factor analyses were conducted using all 2,861 subjects with complete HPS responses (1371 female; age range: 60-82 years; mean age: 70.0). Given the dichotomous nature of the HPS items, a tetrachoric correlation matrix was performed first. Subsequently, minimum residual factoring was carried out followed by promax rotation. The number of factors was determined by scree plot inspection, the Very Simple Structure (VSS) criterion for complexity one and two, and Velicer’s Minimum Average Partial (MAP) criterion. We also assessed the congruency of the factor solution with results of a hierarchical cluster analysis that was carried out with the ICLUST function of the R package psych 3. Factor scores were calculated using the function score.irt.2, which uses the two parameter Item Response Theory (IRT) equivalent of loadings and difficulties. Factor score calculations were based on those items showing discrimination parameters higher than or equal to the cut-off value 0.300.
21
Supplementary Results
Factorial structure of the HPS The scree test suggested a three-factor solution (Fig. S1). The other criteria indicated a one- (VSS for complexity one), three- (VSS for complexity two), and five-factor solution (Velicer MAP). Supplementary Fig. S1. Scree plot showing the eigenvalues of the principal factors derived from the tetrachoric correlation matrix of HPS items
Eigenvalues are based on minimum residual factoring. Note that the eigenvalues of the 3-factor solution do not perfectly correspond to the shown solution.
The hierarchical cluster analysis stopped combinations at three clusters (see Figure S2). As we achieved the highest agreement across the abovementioned indices with the three-factor structure, we decided to proceed with the three-factor model. The sums of squared loadings for the three factors derived from the minimum residual factor analysis were 9.87, 3.14, and 2.47. Following promax rotation, the sums of squared loadings were 7.62, 5.27, and 2.58, accounting for 35% of the variance observed in the 44 items. Figure S3 shows the correlation matrix of the HPS-items which were derived from the factor analysis. Factors one and two were correlated (0.45), while factor three was largely independent from factors one (0.06) and two (-0.05). The congruency of the factor-cluster solution was high, particularly regarding the first factor (.93), while factors two (.80) and three (.71) showed lower congruency coefficients.
22
Supplementary Fig. S2. iCLUST output based on the tetrachoric correlation matrix of HPS items
The two criteria for cluster combinations were as follows: (a) for two clusters of three or more items, combine only if the resulting cluster increases alpha past the maximum of the two subclusters, and (b) for two clusters of four items or more, combine only if the resulting cluster increases beta beyond the maximum of the two. Colored boxes indicate the congruency with factor analysis results, that is, colors indicate for which of the three factors hypomanic core, social vitality, and ordinariness (from dark to light blue) highest factor loadings were obtained.
23
Supplementary Fig. S3. Tetrachoric correlation matrix of the HPS items
Squared factor loadings are shown for each item as blue shaded bars below the correlation matrix.
24
Supplementary Table S1 shows the loadings and IRT parameters of the HPS three-factor model and compares it to the factor solutions reported in two younger non-German samples by Rawlings et al. 4 and Schalet et al. 5 The first factor comprises items with clear hypomanic content and largely overlaps with the factor ‘moodiness’ by Rawlings et al. 4 and the factors ‘mood volatility’ and ‘excitement’ by Schalet and colleagues 5. Note that the term ‘mood volatility’ has been criticized as only very few items actually ask for the lability of subjects’ mood. Terrien et al. 6 therefore named the first factor in their confirmatory factor analysis ‘hypomanic mood’ . However, because the first factor also comprises cognitive and energetic aspects, we use the term Hypomanic Core . Items loading high on the second factor describe high self-confidence, social dominance and leadership. This second factor largely overlaps with those previously labelled as ‘hypersociability’ and ‘social vitality’. In our study, we use the latter term, Social Vitality . The third factor appears to reflect a characteristic that has previously been recognized as ‘ordinariness’, as it describes balanced and controlled people who describe themselves as average persons. 4 Thus, we retain the term Ordinariness . The factor score calculations for Hypomanic Core, Social Vitality and Ordinariness were based on those items showing factor discrimination parameters higher than or equal to the cut-off value 0.300.
25
SuppSuppSupplementarySupp lementary Table S1. Loadings and IRT parameters of the HPS three-factor model IRT IRT Loading discrimination parameter difficulty parameter
Item HYP SOC ORD HYP SOC ORD HYP SOC ORD Rawlings et al. 2000 Schalet et al. 2011 Item text 05 0.4790.4790.479 0.126 0.047 0.5450.5450.545 0.127 0.047 0.0680.0680.068 0.061 0.060 Cognitive Mood Volatility Sometimes ideas and insights come to me so fast that I cannot express them all. 07 0.4080.4080.408 0.227 -0.053 0.4470.4470.447 0.233 -0.053 0.9540.9540.954 0.894 0.872 excluded Social Vitality In unfamil iar surroundings, I am often so assertive and sociable that I surprise myself . . . 08 0.6440.6440.644 -0.196 -0.172 0.8430.8430.843 -0.200 -0.174 0.7190.7190.719 0.561 0.558 Moodiness Mood Volatility There are often times when I am so restless that it is impossible for me to sit still. 10 0.4850.4850.485 0.128 0.160 0.5550.5550.555 0.130 0.162 0.1110.1110.111 0.098 0.098 Moodiness Mood Volatility When I feel an emotion, I usually feel it with extreme intensity. 11 0.5690.5690.569 -0.009 -0.022 0.6920.6920.692 -0.009 -0.022 1.0451.0451.045 0.860 0.860 Moodiness Excitement I am fr equently in such high spirits that I can’t concentrate on any one thing for too . . . 15 0.5220.5220.522 0.015 0.106 0.6120.6120.612 0.015 0.107 0.9320.9320.932 0.795 0.800 Moodiness Excitement I often feel excited and happy for no apparent reason. 18 0.3860.3860.386 0.279 0.045 0.4180.4180.418 0.291 0.045 1.1701.1701.170 1.124 1.081 Moodiness Excitement I often have moods where I feel so energetic and optimistic that I feel I could . . . 19 0.3430.3430.343 0.229 0.042 0.3650.3650.365 0.235 0.042 0.4950.4950.495 0.478 0.466 Cognitive Social Vitality I have such a wide range of i nterests that I often don’t know what to do next. 20 0.4960.4960.496 0.194 -0.034 0.5710.5710.571 0.198 -0.034 1.5811.5811.581 1.399 1.373 Moodiness Mood Volatility There have often been times when I had such an excess of energy that I felt little . . . 22 0.6970.6970.697 0.066 0.057 0.0.0. 972972972 0.066 0.058 0.9530.9530.953 0.685 0.685 Moodiness Mood Volatility I very frequently get into moods where I wish I could be everywhere and do everything . . . 33 0.6200.6200.620 0.125 0.005 0.7890.7890.789 0.126 0.005 1.7041.7041.704 1.348 1.338 Moodiness Excitement I often get so hap py and energetic that I am almost giddy. 35 0.5480.5480.548 -0.010 -0.120 0.6540.6540.654 -0.010 -0.121 1.2571.2571.257 1.052 1.059 Moodiness Mood Volatility I often get into moods where I feel like many of the rules of life don’t apply to me. 38 0.7920.7920.792 -0.202 -0.137 1.2971.2971.297 -0.20 7 -0.139 0.9680.9680.968 0.604 0.597 Moodiness Mood Volatility I frequently find that my thoughts are racing. 39 0.3330.3330.333 0.281 0.039 0.3530.3530.353 0.293 0.039 1.1891.1891.189 1.168 1.122 excluded Social Vitality I am so good at controlling others that it sometimes scares me. 41 0.4000.4000.400 0.159 0.193 0.4370.4370.437 0.161 0.197 0.2690.2690.269 0.250 0.251 Cognitive Mood Volatility I do most of my best work during brief periods of intense inspiration. 43 0.3700.3700.370 0.153 0.125 0.3980.3980.398 0.155 0.126 0.4650.4650.465 0.437 0.436 Cognitive Mood Volatility I have often been so excited about an involving project that I didn’t care about eating . . . 45 0.7630.7630.763 -0.177 -0.168 1.1821.1821.182 -0.179 -0.170 1.5461.5461.546 1.014 1.013 Moodiness Mood Volatility I have often felt happy and irritable at the same time. 46 0.7040.7040.704 -0.118 -0.114 0.9920.9920.992 -0.118 -0.115 1.9921.9921.992 1.424 1.424 Moodiness Excitement I often get into excited moods where it’s almost impossible for me to stop talking. 44 0.8280.8280.828 -0.258 -0.293 1.4761.4761.476 -0.267 ---0.3070.3070.307 1.8511.8511.851 1.074 1.0861.0861.086 Moodiness Mood Volatility I frequently get i nto moods where I feel very speeded -up and irritable. 37 0.7470.7470.747 -0.262 ---0.3180.3180.318 1.1231.1231.123 -0.272 ---0.3350.3350.335 1.0381.0381.038 0.715 0.7280.7280.728 Moodiness Mood Volatility I seem to be a person whose mood goes up and down easily. 32 0.4450.4450.445 0.3120.3120.312 ---0.4150.4150.415 0.4960.4960.496 0.3280.3280.328 ---0.4560.4560.456 1.7891.7891.789 111...686686686 1.7611.7611.761 Hypersociability Excitement I am considered to be kind of a “hyper” person. 09 0.3340.3340.334 0.3740.3740.374 -0.162 0.3550.3550.355 0.4030.4030.403 -0.164 0.9960.9960.996 1.0121.0121.012 0.951 excluded Mood Volatility Many people consider me to be amusing but kind of eccentric. 02 0.135 ---0.5350.5350.535 0.113 0.137 ---0.6330.6330.633 0.114 -0.955 ---1.1201.1201.120 -0.953 Hypersociability Social Vitality It would make me nervous to play the clown in front of other people. 04 0.029 0.5700.5700.570 0.019 0.029 0.6930.6930.693 0.019 1.706 2.0752.0752.075 1.706 Hypersociability Social Vitality I think I would make a good nightclub comedian. 06 0.093 ---0.4560.4560.456 0.192 0.094 ---0.5130.5130.513 0.196 -0.657 ---0.7350.7350.735 -0.666 Hypersociability Social Vitality When with groups of people, I usually prefer to let someone else be the center . . . 13 0.104 0.4960.4960.496 0.265 0.105 0.0.0. 572572572 0.275 -0.068 ---0.0780.0780.078 -0.070 Cognitive Social Vitality People often come to me when they need a clever idea. 25 0.182 ---0.5240.5240.524 0.039 0.185 ---0.6150.6150.615 0.039 -0.731 ---0.8430.8430.843 -0.719 Hypersociability Social Vitality When I go to a gathering where I don’t know anyone, it usually takes me a while . . . 26 0.042 0.5740.5740.574 -0.049 0.042 0.7020.7020.702 -0.049 0.954 1.1651.1651.165 0.954 excluded Social Vitality I think I would make a good actor, because I can play many roles convincingly. 28 0.193 0.293 0.017 0.196 0.3060.3060.306 0.017 1.155 1.1851.1851.185 1.133 Cognitive Excluded I frequently write down the thoughts and insights that come to me when I am thinking . . . 29 0.244 0.4690.4690.469 0.031 0.252 0.5310.5310.531 0.031 1.196 1.3131.3131.313 1.161 Hypersociability Social Vitality I have often persuaded groups of friends to do something really adventurous or crazy. 30 0.005 0.5150.5150.515 -0.027 0.005 0.6010.6010.601 -0.027 1.282 1.4951.4951.495 1.282 excluded Social Vitality I would really enjoy being a politician and hitting the campaign trail. 36 0.073 0.6520.6520.652 0.120 0.073 0.8590.8590.859 0.120 0.518 0.6810.6810.681 0.520 excluded Social Vitality I find it easy to get others to become sexually interested in me. a 40 0.181 0.5730.5730.573 0.056 0.184 0.6990.6990.699 0.056 1.181 1.4171.4171.417 1.164 Hypersociability Social Vitality At social gatherings, I am usually the “life of the party.” 42 0.218 0.6210.6210.621 0.086 0.224 0.7930.7930.793 0.086 0.990 1.2321.2321.232 0.969 Cognitive Social Vitality I seem to have an uncommon ability to persuade and inspire others. 01 0.218 ---0.6440.6440.644 0.3070.3070.307 0.224 ---0.8420.8420.842 0.3230.3230.323 -1.516 ---1.9341.9341.934 ---1.5541.5541.554 Ordinariness Social Vi tality I consider myself to be pretty much an average kind of person. 14 0.079 ---0.3190.3190.319 0.3300.3300.330 0.080 ---0.3370.3370.337 0.3490.3490.349 -0.973 ---1.0241.0241.024 ---1.0271.0271.027 Ordinariness Social Vitality I am no more self -aware than the majority of people. 27 0.123 ---0.4420.4420.442 0.5170.5170.517 0.124 ---0.4920.4920.492 0.6050.6050.605 -2.003 ---2.2162.2162.216 ---2.3232.3232.323 Ordinariness Social Vitality I like to have others think of me as a normal kind of person. 12 0.218 0.164 0.3170.3170.317 0.223 0.166 0.3340.3340.334 -0.383 -0.379 ---0.3940.3940.394 Cognitive Excluded I sometimes have felt that nothing can happen to me until I do what I am meant to . . . 21 -0.180 -0.107 0.5590.5590.559 -0.184 -0.108 0.6740.6740.674 -1.243 -1.229 ---1.4741.4741.474 Moodiness Mood Volatility My moods do not seem to fluctuate any more than most people’s do. 24 -0.057 0.154 0.3770.3770.377 -0.057 0.155 0.4070.4070.407 -1.577 -1.59 4 ---1.7001.7001.700 Moodiness Excluded When I feel very excited and happy, I almost always know the reason why. 31 -0.126 0.030 0.4650.4650.465 -0.127 0.030 0.5260.5260.526 -2.023 -2.008 ---2.2672.2672.267 Moodiness Mood Volatility I can usually slow myself down when I want to. 17 ---0.3140.3140.314 0. 100 0.7090.7090.709 ---0.3300.3300.330 0.101 1.0061.0061.006 ---1.6091.6091.609 -1.536 ---2.1682.1682.168 excluded Excitement I am usually in an average sort of mood, not too high and not too low. 16 -0.050 -0.156 0.241 -0.050 -0.158 0.248 -0.431 -0.436 -0.443 Ordinariness Social Vitality I can’t imagin e that anyone would ever write a book about my life. 47 0.181 -0.228 0.268 0.184 -0.234 0.278 -0.891 -0.900 -0.909 Ordinariness Social Vitality I would rather be an ordinary success in life than a spectacular failure. HPS: Hypomanic Personality Scale (Eckblad & Chapman, 1986), IRT: Item response theory, HYP: HPS facet ‘hypomanic core‘, SOC: HPS facet ‘social vitality’, ORD: HPS facet ‘ordinariness’, a No sexual content in the German translation 1 of the item. Minimum residual factor analysis was based on tetrachoric correlations. Loadings and IRT discrimination parameters are bold where they are higher than or equal to the cut-off value 0.300. IRT difficulty parameters are bold where the corresponding IRT discrimination parameters are higher than or equal to the cut-off value 0.300. Note that subsequent factor score calculations were based on those items showing factor discrimination parameters higher than or equal to the cut-off value 0.300.
26
Correlation analyses Extending from meta-analyses’ findings showing various forms of sleep impairment in euthymic BD 7,8 , the current study examined the association of individual sleep variables with the HPS, a psychometric risk factor for BD. Because each test was hypothesis-driven, there was no need for multiple-testing correction 9,10 . However, to confirm that results persisted after correction for multiple testing, we additionally calculated the False Discovery Rate (FDR)-corrected p values according to Benjamini-Hochberg 11 and regarded associations with FDR < 0.05 as significant after multiple testing correction. In total, 40 out of the 46 nominal significant correlations remained significant after multiple-testing correction (see Table S2).
Supplementary Table S2. Partial Spearman correlations between HPS and sleep-wake variables. FDR- corrected and nominal p-values are given for comparison HPS total HPS subscale HPS subscale HPS subscale
sum-score Hypomanic Core Social Vitality Ordinariness
rho p FDR rho p FDR rho p FDR rho p FDR
Actigraphy (n = 771) Means Sleep-onset latency .038 .294 .399 .042 .240 .342 .054 .134 .200 -.035 .329 .432 Sleep-onset time -.006 .864 .907 -.035 .329 .432 .025 .497 .580 .061 .089 .144 Sleep-offset time -.013 .721 .797 -.041 .252 .353 .007 .851 .905 .056 .118 .180 Sleep duration -.079 .029 .057 * -.072 .046 .084 * -.059 .101 .160 .034 .350 .439 NWAK .078 .030 .058 * .069 .056 .098 .069 .056 .098 -.011 .757 .826 WASO .103 .004 .014 ** .091 .011 .025 ** .080 .027 .055 * -.046 .203 .299 Sleep efficiency -.106 .003 .011 ** -.101 .005 .016 ** -.086 .017 .035 ** .033 .361 .446
NightNight----totototo----nightnight variability Sleep-onset latency .019 .594 .665 -.005 .893 .926 .076 .036 .067 * -.026 .470 .556 Sleep-onset time .122 7E-4 .005 ** .102 .004 .014 ** .113 .002 .010 ** .009 .813 .876 Sleep-offset time .110 .002 .010 ** .119 9E-4 .006 ** .069 .057 .098 -.062 .087 .144 Sleep duration .098 .006 .018 ** .092 .011 .025 ** .097 .007 .019 ** -.003 .931 .943 NWAK .145 6E-5 9E-4** .109 .002 .010 ** .152 2E-5 7E-4** -.092 .011 .025 ** WASO .115 .001 .007 ** .116 .001 .007 ** .095 .008 .020 ** -.109 .002 .010 ** Sleep efficiency .106 .003 .011 ** .095 .008 .020 ** .103 .004 .014 ** -.092 .011 .025 **
PSQI (n = 1766) Sleep-onset latencya .013 .576 .660 .078 .001 .007 ** -.080 8E-4 .005 ** -.098 4E-5 8E-4** Bedtime a,b .029 .230 .333 .018 .444 .533 .041 .087 .144 .001 .968 .968 Get-up time a -.022 .348 .439 -.023 .344 .439 -.020 .413 .502 -.013 .581 .660 Sleep durationa -.038 .114 .177 -.095 7E-5 9E-4** .027 .261 .359 .081 6E-4 .005 ** Sleep efficiency c .002 .925 .943 -.065 .006 .018 ** .071 .003 .011 ** .091 1E-4 .002 ** Daytime sleepiness a .075 .002 .010 ** .088 2E-4 .002 ** .052 .028 .057 * -.086 3E-4 .003 ** PSQI score d .063 .008 .020 ** .158 3E-11 1E-9** -.063 .008 .020 ** -.165 3E-12 3E-10 ** Night-to-night variability is operationalized by intraindividual standard deviation (ISD) across a single subject's multiple nights. Note that Hypomanic Core, Social Vitality and Ordinariness here refer to factor scores derived from factor analyses. Results were additionally confirmed by analyses with traditional sum scores (see Supplementary Table 4). Effects of sex and age were partialled out. FDR: False Discovery Rate according to Benjamini and Hochberg 11 ; NWAK: Number of awakenings; WASO: wake after sleep-onset time * p < .05 (two-sided nominal significance) ** FDR < .05 (p value corrected for all tested associations applying the Benjamini-Hochberg FDR method 11 ) a based on the respective PSQI item b time subject goes to bed; c quotient of sleep duration and time in bed with the latter calculated from bedtime and get-up time; d PSQI total score calculated according the manual from all PSQI components; higher PSQI scores mean worse sleep quality.
27
Supplementary Table S3. Comparison of Spearman correlations
Spearman’s rho P value of Spearman’s rho comparisons
HT vs. HT vs. HT vs. HC vs. HC vs. SV vs. HT HC SV OD HC SV OD SV OD OD
Actigraphy (n = 771) Means Sleep-onset latency .038 .042 .054 -.035 .831 .549 .211 .779 .180 .112 Sleep-onset time -.006 -.035 .025 .061 .172 .258 .247 .153 .095 .512 Sleep-offset time -.013 -.041 .007 .056 .181 .469 .2 35 .249 .091 .377 Sleep duration -.079 -.072 -.059 .034 .750 .471 .054 .759 .068 .098 NWAK .078 .069 .069 -.011 .667 .734 .125 .999 .165 .153 WASO .103 .091 .080 -.046 .593 .399 .011 ** .782 .018* .025 * Sleep efficiency -.106 -.101 -.086 .033 .797 .467 .017 * .732 .021* .033 * NightNight----totototo----nightnight variability Sleep-onset latency .019 -.005 .076 -.026 .257 .038* .437 .054 .714 .070 Sleep-onset time .122 .102 .113 .009 .357 .727 .051 .808 .104 .063 Sleep-offset time .110 .119 .069 -.062 .680 .124 .003 ** .226 .002** .020 * Sleep duration .098 .092 .097 -.003 .748 .963 .081 .894 .101 .074 NWAK .145 .109 .152 -.092 .092 .787 5E-5** .301 5E-4** 1E-5** WASO .115 .116 .095 -.109 .963 .463 1E-4** .616 9E-5** 3E-4** Sleep efficiency .106 .095 .103 -.092 .599 .923 7E-4** .838 .001** 5E-4** PSQI (n = 1766) Sleep-onset latency a .013 .078 -.080 -.098 2E-6** 1E-7** .004 ** 5E-9** 4E-6** .624 Bedtime a,b .029 .018 .041 .001 .450 .493 .470 .408 .653 .273 Get-up time a -.022 -.023 -.020 -.013 .989 .872 .810 .911 .806 .860 Sleep duration a -.038 -.095 .027 .081 3E-5** 3E-4** .002 ** 7E-6** 4E-6** .132 Sleep efficiency c .002 -.065 .071 .091 1E-6** 1E-4** .021 * 6E-7** 5E-5** .586 Daytime sleepiness a .075 .088 .052 -.086 .315 .202 3E-5** .181 5E-6** 1E-4** 7E- 2E- 2E- PSQI score d .063 .158 -.063 -.165 5E-12** ** 2E-9** ** ** .004 ** 13 16 17 Comparisons of Spearman’s rho correlation coefficients was carried out according to the formula by Dunn & Clark 12 as implemented in R package cocor v.1.1-313 . HT HPS Total ; HC Hypomanic Core ; SV Social Vitality ; OD Ordinariness * p < .05 (two-sided nominal significance) ** FDR < .05 (p value corrected for all tested associations using Benjamini-Hochberg FDR method) Night-to-night variability is operationalized by intraindividual standard deviation (ISD) across a single subject's multiple nights. a based on the respective PSQI item b time subject goes to bed; c quotient of sleep duration and time in bed with the latter calculated from bedtime and get-up time; d PSQI total score calculated according the m anual from all PSQI components; higher PSQI scores mean worse sleep quality.
28
Secondary analyses for confirmation of results
Further analyses were conducted to confirm the correlational results. First, all correlational analyses were repeated with unweighted sum scores for each HPS subscale instead of factor scores, which resulted in comparable, albeit, as expected, somewhat weaker associations (see Table S1). In total, 39 out of 84 correlations reached the level of nominal significance. Of those, 32 remained significant after multiple- testing correction. A reduction in the number of significant associations was most pronounced for the facet Ordinariness .
Supplementary Table S4. Partial Spearman correlations between HPS (unweighted sum scores) and the sleep- wake variables
Unweighted sum score Unweighted sum score Unweighted sum score HPS sum score of of of Hypomanic Core Social Vitality Ordinariness rho p FDR rho p FDR rho p FDR rho p FDR
Actigraphy (n = 771) Means Sleep-onset latency .038 .294 .419 .032 .372 .480 .056 .124 .200 -.013 .719 .784 Sleep-onset time -.006 .864 .874 -.008 .816 .836 .044 .228 .339 .033 .356 .475 Sleep-offset time -.013 .721 .784 -.024 .508 .611 .013 .728 .784 .045 .214 .326 Sleep duration -.079 .029 .068* -.079 .029 .068 * -.063 .083 .151 .024 .509 .611 NWAK .078 .030 .068* .063 .080 .150 .059 .101 .170 .027 .456 .563 WASO .103 .004 .020** .085 .018 .047 ** .069 .056 .110 .009 .812 .836 Sleep efficiency -.106 .003 .015** -.094 .009 .030 ** -.079 .029 .068* -.010 .775 .818 Night ---tototo ---night
variability Sleep-onset latency .019 .594 .674 -.010 .779 .818 .075 .038 .083* -.033 .366 .480 Sleep-onset time .122 7E-4 .007** .098 .006 .025 ** .123 7E-4 .007** .077 .033 .073 * Sleep-offset time .110 .002 .012** .125 5E-4 .007 ** .059 .100 .170 -.028 .436 .547 Sleep duration .098 .006 .025** .087 .015 .043 ** .097 .007 .027** .061 .093 .16 7 NWAK .145 6E-5 .002** .115 .001 .010 ** .145 6E-5 .002** -.022 .535 .633 WASO .115 .001 .010** .107 .003 .015 ** .082 .024 .060* -.047 .193 .304 Sleep efficiency .106 .003 .015** .090 .013 .037 ** .094 .009 .030** -.036 .315 .434 PPPSQIPSQISQISQI (n = 1766) Sleep -onset .013 .576 .663 .063 .008 .029 ** -.079 8E-4 .007** -.057 .017 .045 ** latency a Bedtime a,b .029 .230 .339 .031 .196 .304 .046 .051 .105 .009 .699 .783 Get-up time a -.022 .348 .471 -.025 .299 .419 -.019 .432 .547 -.013 .573 .663 Sleep duration a -.038 .114 .187 -.088 2E-4 .004 ** .028 .236 .342 .040 .096 .168 Sleep efficiency c .002 .925 .925 -.046 .053 .105 .073 .002 .012** .046 .051 .105 Daytime sleepiness a .075 .002 .012** .081 6E-4 .007 ** .061 .010 .031** -.042 .080 .150 PSQI score d .063 .008 .029** .127 9E-8 7E-6 ** -.061 .011 .032** -.088 2E-4 .004 ** Effects of sex and age were partialled out. FDR: False Discovery Rate according to Benjamini and Hochberg 11 ; NWAK: Number of awakenings; WASO: wake after sleep-onset time * p < .05 (two-sided nominal significance) ** FDR < .05 (p value corrected for all tested associations applying the Benjamini-Hochberg FDR method 11 ) Night-to-night variability is operationalized by intraindividual standard deviation (ISD) across a single subject's multiple nights. a based on the respective PSQI item b time subject goes to bed; c quotient of sleep duration and time in bed with the latter calculated from bedtime and get-up time; d PSQI total score calculated according the manual from all PSQI components; higher PSQI scores mean worse sleep quality.
29
As another confirmation of results, analyses were repeated using HPS decile extreme groups (as is often done in psychometric high-risk studies). HPS extreme groups were defined as the lower (HPS-); and upper (HPS+) decile of the distribution (HPS- : n=61, range 0-2, mean sum score = 1.39, SD=0.74; HPS +: n=63, mean sum score = 20.23, SD = 3.79, range: 16-37). Results from extreme group comparisons are presented in Supplementary Table S5 and illustrated in Supplementary Figure S4. In total, 38 out of 84 comparisons reached the level of nominal significance, of which 23 remained significant after multiple-testing correction. Concerning objective sleep, HPS high-scorers had significantly lower sleep efficiency and more time WASO. In accordance with the correlational analyses, the extreme groups more strongly differed concerning the night-to-night variability of sleep parameters than concerning the mean sleep parameters. The HPS high scorers showed greater variability in all sleep variables, except for of sleep-onset latency which did not reach significance level. Also in line with correlation analyses, HPS high-scorers had lower self-reported sleep efficiency, worse overall sleep quality (PSQI total score) and greater daytime sleepiness.
30
Supplementary Table S5S5. HPS total sum score and sleep-wake variables – extreme group comparisons
Inference statistics
HPS+ HPS- Mean ranks a χ² η² p FDR
N 63 61
Demography Sex (f/m) 28 / 35 27 / 34 0.000 < 0.001 .984 Age (yrs) 70.2 (4.1) 70.1 (4.2) 63.12 / 61.86 0.038 < 0.001 .845
Actigraphy Means Sleep-onset latency (min) 0:09 (0:06) 0:07 (0:04) 68.13 / 56.69 3.140 0.025 .076 .140 Sleep-onset time (h:min) 23:29 (0:51) 23:29 (0:49) 61.58 / 63.45 0.084 0.001 .772 .809 Sleep-offset time (h:min) 7:00 (0:43) 7:04 (0:44) 62.06 / 62.96 0.020 < 0.001 .889 .889 Sleep duration (h:min) 6:12 (1:15) 6:34 (0:53) 58.78 / 66.34 1.374 0.011 .241 .332 NWAK 3.23 (1.2) 2.92 (1.2) 66.30 / 58.57 1.436 0.012 .231 .332 WASO (h:min) 1:19 (0:39) 1:00 (0:29) 70.83 / 53.90 6.872 0.055 .009 .021 ** Sleep efficiency (%) 81.8 (6.3) 85.3 (6.2) 53.05 / 72.26 8.859 0.071 .003 .011 **
NightNightNight-Night ---totototo----nightnight variability Sleep-onset latency (min) 0:07 (0:04) 0:06 (0:03) 65.84 / 59.05 1.107 0.009 .293 .379 Sleep-onset time (h:min) 0:46 (0:24) 0:35 (0:19) 70.13 / 54.62 5.768 0.047 .016 .036 ** Sleep-offset time (h:min) 0:52 (0:35) 0:37 (0:18) 73.52 / 51.11 12.049 0.097 5E-4 .003 ** Sleep duration (h:min) 0:59 (0:20) 0:46 (0:18) 73.83 / 50.80 12.717 0.103 4E-4 .003 ** NWAK 1.62 (0.5) 1.29 (0.5) 74.49 / 50.11 14.267 0.115 2E-4 .003 ** WASO (h:min) 0:39 (0:21) 0:27 (0:14) 73.32 / 51.33 11.602 0.094 7E-4 .003 ** Sleep efficiency (%) 6.73 (2.6) 5.50 (2.2) 71.14 / 53.57 7.406 0.060 .006 .018 **
PSQIPSQIPSQI Sleep-onset latency b 0:24 (0:26) 0:16 (0:14) 66.31 / 58.57 1.485 0.012 .223 .332 Bedtime b,c 23:00 (0:52) 22:52 (0:46) 65.34 / 59.57 0.841 0.007 .359 .439 Get-up time b 7:24 (0:45) 7:16 (0:42) 64.79 / 60.13 0.545 0.004 .460 .533 Sleep duration b 6:51 (1:05) 7:09 (1:03) 57.54 / 67.62 2.714 0.022 .0 99 .168 Sleep efficiency d 82.1 (11.9) 85.5 (10.6) 56.60 / 68.60 3.475 0.028 .062 .125 Daytime sleepiness b 0.30 (0.64) 0.07 (0.31) 67.77 / 57.06 7.722 0.062 .005 .017 ** PSQI score e 5.87 (3.2) 3.92 (2.9) 74.05 / 50.57 13.444 0.108 2E-4 .003 **
Except for sex and n, descriptive statistics are presented as mean (standard deviation). Inference statistics are based on χ²-Test (sex), or Kruskal-Wallis Test (all other variables). All χ² values are specified by one degree of freedom. The effect size η² was calculated by squaring r derived from r=√(χ²/N) according to Rosenthal & DiMatteo 14 . FDR: False Discovery Rate according to Benjamini and Hochberg 11 * p < .05 (two-sided nominal significance) ** FDR < .05 (p value corrected for all shown associations applying Benjamini-Hochberg FDR method 11 ) Night-to-night variability is operationalized by intraindividual standard deviation (ISD) across a single subject's multiple nights. a Mean rank HPS+ / mean rank HPS- b based on the respective PSQI item c time subject goes to bed; d quotient of sleep duration and time in bed with the latter calculated from bedtime and get-up time; e PSQI total score calculated according the manual from all PSQI components; higher PSQI scores mean worse sleep quality.
31
Supplementary Fig. S4. Boxplots of actigraphy and PSQI variables stratified by HPS extreme groups
Intraindividual night-to-night variability is operationalized by intraindividual standard deviation (ISD) across a single subject's multiple nights. Boxplots are stratified by top and bottom decile HPS groups (HPS+, N=63 vs. HPS-, N=61). Boxes represent the interquartile range of each distribution (data between the lower and upper quartile), with the horizontal line corresponding to the median. Whiskers extend to the furthest observation within 1.5 times the interquartile range from the lower and upper quartile. Dots represent single data points, jittered horizontally to avoid overplotting. Note that statistical analyses were based on nonparametric (ranked) data.
32
Exploratory regression analysis with HPS subscales In the present study, lowest type-I-error probabilities were obtained for associations between the PSQI total score (an indicator for poor sleep quality) and higher scores on the Hypomanic Core subscale (rho = .158, P = 3E-11), and lower scores on the Ordinariness subscale (rho = -.165, P = 3E-12). Although correlations with Social Vitality were weaker (rho = -.063, P = .008), an exploratory linear regression analysis with ranked variables (resembling Spearman correlations) revealed that all three subscales explained independent variance in PSQI total scores. In total, 6.0% of the variance in PSQI total scores (adjusted for sex and age) was explained by the weighted combination of Hypomanic Core (β = .161, P = 1.3E-10, ηp² = 0.023), Social Vitality (β = -.139, P = 8.6E-9, ηp² = 0.019), and Ordinariness (β = -.135, P = 1.7E-8, ηp² = 0.018). The strength of this association was substantial when compared to using the HPS total score ( HPS
Total ) as the regressor, which resulted in an explained variance of only 0.4% ( ηp² = 0.004, P = .008). The observed failure of the HPS Total appeared to be attributable to the facet Social Vitality. Whereas the facet Hypomanic Core positively correlated with both HPS Total (rho = .832) and poor sleep quality and Ordinariness negatively correlated with both the HPS Total (rho = -.292) and poor sleep quality, Social Vitality correlated positively with HPS Total (rho = .721) but negatively with poor sleep quality. Hence, the multidimensional HPS scale differentially associates with subjective sleep quality, with associations of aggregated scales apparently cancelling each other out. To unravel the mechanisms between hypomanic personality and sleep, the differentiation of these facets is recommended.
33
Prediction of HPS by sleep and covariates To investigate the relative importance of objective and subjective sleep variables and covariates in predicting Hypomanic Personality Scores, we carried out a series of regression analyses. In order to avoid overfitting and inflated R 2 estimates, we used a tenfold cross-validation approach which involves evaluating the performance of a prediction model by applying the model to new data not used in training it. We selected N = 640 subjects with valid data from both the actigraphy assessment and the PSQI questionnaire. From the set of 21 sleep variables, we dropped the actigraphy variable ‘mean sleep-offset time’ and the PSQI variable ‘sleep-efficiency’ due to reasons of multicollinearity (stepwise exclusion of variables with a variance inflation factor > 10). We sought to determine the incremental predictive value of sleep variables by comparing the performance of a full model to the performance of a standard covariate model. The standard covariate model contained the predictors sex, age, socioeconomic status (SES), body mass index (BMI), and questionnaire-derived indicators of sleep apnea and Periodic Limb Movement Disorder (PLMD). We selected these variables due to their extensive implications for sleep behavior and affective traits. Socioeconomic status was calculated as a combination of education, working status, and household income as previously described. 15 Information on sleep apnea and PLMD were drawn from two PSQI items that were answered by the participant’s partner and were not aggregated to the PSQI total score (‘long pauses between breaths while asleep’, ‘legs twitching or jerking while asleep’). These two PSQI items were rated on a 4-point scale. Missing item responses (22% and 20%, respectively) were substituted by the sample median.
We analyzed the performance of three different full models. The first full model contained the actigraphy variables, while the second model contained the PSQI variables. The third full model included both the actigraphy and PSQI variables. All full models included the variables of the standard covariate model as described above.
In order to carry out the tenfold cross-validation procedure, our sample was split into ten equal sized subsets. To ensure comparable distributions of HPS scores across the ten subsets, the sample was sorted according to the respective HPS score (total or subscale score) and subjects of the first, second, third, …, 64th decuplet were each randomly assigned to one of the ten subsets. Next, one of the subsets was selected as the testing dataset, while the other nine subsets served as the training dataset. After the first model was trained and tested, the next subset was selected as the testing sample, while the other nine subsets served as the training sample. This procedure was carried on until each subset served exactly once as the testing dataset. The tenfold cross-validation procedure with random group assignment was repeated 100 times so that 100 predictions were made for each subject. Predicted HPS scores were then averaged and correlated with the measured HPS scores.
Prediction models were built using the gradient boosting package ‘xgboost’ (v.0.82.1) 16 in R. The xgboost algorithms have been widely used in the field of supervised machine learning and their solutions have been awarded in several data science competitions. We trained our models with the tree booster. The learning rate was set to eta = 0.02 with a maximum tree depth of 3. Training iterations were set to 1000. Of the nine training subsets, eight subsets served to fit the model, while one subset was selected to monitor model performance and avoid overfitting. The training process was stopped after 50 iterations if no further improvement was achieved. We used default settings for all other training parameters.
Supplementary Table S6 shows the Pearson correlations between predicted and measured HPS scores for the standard covariate model and the three full models. ∆R2 reflects the increase in explained variance of a full model relative to the standard covariate model and was calculated as follows:
2 2 2 ∆R = rho Full - rho Standard Significance of ∆R2 was calculated by comparing the correlations of the full and standard model using the formula by Dunn & Clark 12 as implemented in the R package cocor v.1.1-313 .
34
Supplementary Table S6S6. Pearson correlations between predicted and measured HPS scores
HPS total HPS subscale HPS subscale HPS subscale
sum-score Hypomanic Core Social Vitality Ordinariness
rho p ∆R2 rho p ∆R2 rho p ∆R2 rho p ∆R2 Standard model .015 .696 -.002 .954 .166 2E-5 -.034 .388 Full model acti .111 .005 0.012* .145 2E-4 0.021* .162 4E-5 -0.001 .087 .028 0.006* Full model PSQI .075 .057 0.005 .179 5E-6 0.032** .141 3E-4 -0.008 .066 .095 0.003* Full model acti + PSQI .135 .001 0.018* .246 3E-10 0.060** .157 7E-5 -0.003 .113 .004 0.012* Predicted HPS scores were derived from a tenfold cross-validation procedure with 100 repeats. Models were trained u sing R package xgboost with booster type ‘gbtree’. ∆R2 reflects the increase in R 2 relative to the standard model. Significance of ∆R2 is indicated by asterisks and was calculated by comparing the correlations of the full and standard model using the formula by Dunn & Clark 12 as implemented in R package cocor v.1.1-313 . * p(∆R2) < .050 (two-sided nominal significance of ∆R2) ** p(∆R2) < .001 (two-sided nominal significance of ∆R2)
Overall, adding objective and subjective sleep variables as features to the machine learning model significantly increased accuracy in predicting HPS scores. The highest predictive accuracy was achieved for the HPS subscale Hypomanic Core , resulting in a Pearson correlation between predicted and measured scores of rho = .246 ( p = 3E-10) and an incremental value of sleep variables of ∆R2 = 0.060 ( p = 7E-7). Furthermore, there was an incremental value of sleep variables in predicting both the subscale Ordinariness (∆R2 = 0.012, p = .003) and the HPS total score ( ∆R2 = 0.018, p = .010). In comparison, although predicted scores of the HPS subscale Social Vitality significantly correlated with the measured scores (rho = .157, p = 7E-5), analyses did not reveal evidence for an incremental value of sleep variables ( ∆R2 = - 0.003, p = .784). In sum, our analyses revealed a predictive value of sleep variables for three of the four HPS scores, with the highest model performance observed for the subscale Hypomanic Core .
Next, we extracted the relative contribution of each variable to the prediction models. We used the function ‘xgb.importance’ in the R package xgboost and averaged the relative feature gain across all full models that included both the actigraphy and PSQI variables. Supplementary Figure S5 shows the relative contribution of each variable to the respective prediction model.
SupplemSupplementaryentary Fig. S5. Bar plot showing the relative importance of each variable to the prediction model
The bar plot illustrates the relative contribution of each feature to the tree-based prediction models. Note that the importance metric allows comparing the role of variables within each model, but does not enable inferences on absolute magnitudes across models. This implies that the bar plot does not account for varying prediction performances. BMI: Body Mass Index; SES: socioeconomic status; cov: covariate; acti: sleep variable from actigraphy; psqi: sleep variable from questionnaire Pittsburgh Sleep Quality Index; NWAK: number of awakenings; WASO: wake after sleep-onset time.
35
SSSupplementarySupplementary References 1. Meyer, T. D., Drüke B. & Hautzinger M. Hypomane Persönlichkeit-Psychometrische Evaluation und erste Ergebnisse zur Validität der deutschen Version der Chapman-Skala. Zeitschrift für Klinische Psychologie und Psychotherapie 292929,29 35-42 (2000). 2. Eckblad, M. & Chapman L. J. Development and validation of a scale for hypomanic personality. J Abnorm Psychol 959595,95 214-222 (1986). 3. Revelle, W. psych: Procedures for Personality and Psychological Research. Northwestern University, Evanston, Illinois, USA. (2017). 4. Rawlings, D., Barrantes-Vidal N., Claridge G., McCreery C. & Galanos G. A factor analytic study of the Hypomanic Personality Scale in British, Spanish and Australian samples. Personality and Individual Differences 282828,28 73-84 (2000). 5. Schalet, B. D., Durbin C. E. & Revelle W. Multidimensional structure of the Hypomanic Personality Scale. Psychol Assess 232323,23 504-522 (2011). 6. Terrien, S., Stefaniak N., Morvan Y. & Besche-Richard C. Factor structure of the French version of the Hypomanic Personality Scale (HPS) in non-clinical young adults. Compr Psychiatry 626262,62 105-113 (2015). 7. Geoffroy, P. A. et al. Sleep in patients with remitted bipolar disorders: a meta-analysis of actigraphy studies. Acta Psychiatr Scand 131131131,131 89-99 (2015). 8. Ng, T. H. et al. Sleep-wake disturbance in interepisode bipolar disorder and high-risk individuals: a systematic review and meta-analysis. Sleep Med Rev 202020,20 46-58 (2015). 9. Morgan, J. F. p Value fetishism and use of the Bonferroni adjustment. Evid Based Ment Health 101010,10 34-35 (2007). 10. Perneger, T. V. Adjusting for multiple testing in studies is less important than other concerns. Bmj 318318318,318 1288 (1999). 11. Benjamini, Y. & Hochberg Y. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society. Series B (Methodological) 575757,57 289-300 (1995). 12. Dunn, O. J. & Clark V. Correlation Coefficients Measured on the Same Individuals. Journal of the American Statistical Association 646464,64 366-377 (1969). 13. Diedenhofen, B. & Musch J. cocor: a comprehensive solution for the statistical comparison of correlations. PLoS ONE 101010,10 e0121945 (2015). 14. Rosenthal, R. & DiMatteo M. R. Meta-analysis: recent developments in quantitative methods for literature reviews. Annu Rev Psychol 525252,52 59-82 (2001). 15. Lampert, T., Kroll L. E., Müters S. & Stolzenberg H. Measurement of socioeconomic status in the German Health Interview and Examination Survey for Adults (DEGS1). Bundesgesundheitsblatt 565656,56 (2013). 16. Chen, T. et al. xgboost: Extreme Gradient Boosting. R package version 0.82.1. https://CRAN.R-project.org/package=xgboost (2019).
36
III)III)III) Summary
Dissertation zur Erlangung des akademischen Grades
Dr. med.
Hypomanic Personality and Sleep eingereicht von: David Wozniak
geboren am 29.07.1990 in Berlin angefertigt an der: Universität Leipzig
Klinik für Psychiatrie und Psychotherapie
Betreuer: Prof. Dr. med. Georg Schomerus, Dr. rer. nat. Tilman Hensch
Prof. Dr. med. Ulrich Hegerl
Dezember, 2019
______
Objectives: The current study analyses the associations of objective and subjective sleep parameters and day- time sleepiness with a risk factor for bipolar disorder (BD), the Hypomanic Personality Scale (HPS). In a large cohort of healthy subjects, we expect associations between the HPS and shorter and more disturbed sleep, increased daytime sleepiness and more night-to-night sleep-variability. In contrast to most prior research, we did not only associate sleep with the HPS total score, but also with HPS subscales. To this end, we conducted the first factor analysis on the German translation of the HPS. We hypothesize that the associations of the HPS with sleep variables will differ with respect to the HPS subscales.
Introduction: Sleep in Bipolar Disorder Sleep disturbances and increased daytime sleepiness occur in BD during manic episodes, but also in the euthymic state 1,2 . Accumulating data suggests that impaired or reduced sleep and increased daytime sleepiness are not only symptoms of BD, but contribute to the disease process itself and to manic and hypomanic behavior in particular 3-9. Experimental and longitudinal studies of patients, healthy subjects and animals suggest that sleep deprivation can induce mania and hypomania 3–9.
37
Further, in a recent study 10 daytime sleepiness predicted manic and hypomanic relapse. Present findings on impaired sleep in euthymic BD leave it unclear to what extent these disturbances are a predisposition, a consequence of the disease process or due to medication. Studying healthy sub- jects who are vulnerable to BD can improve our understanding of whether sleep impairment is a predisposing factor. Genetic high-risk studies have mostly shown that sleep in healthy offspring of BD patients is indeed impaired 11–13 . There is only one longitudinal study, which was not based on genetic high-risk offspring but on a healthy community sample, and which could nonetheless pre- dict subsequent development of BD during a ten-year follow up by poor sleep quality at baseline 14 . Moreover, this prediction of BD by poor sleep persisted when excluding subjects with a family histo- ry of affective disorders. Lack of data and a low probability of incident BD in healthy subjects not genetically at risk are inherent problems of such longitudinal cohort studies. Thus, more data are needed.
The Hypomanic Personality Scale The psychometric high-risk approach 15 , which assesses e.g. BD risk in large populations via ques- tionnaires such as the Hypomanic Personality Scale 16 , is one useful method to detect vulnerable subjects. The HPS as a dimensional scale allows not only for the identification of unusually high scorers (high risk subjects), but also the assessment of the total normal variance. The association of the entire continuum of bipolarity with sleep disturbances can then be examined. The HPS assesses features of hypomanic episodes as described within the diagnostic systems, but also includes corre- lated traits such as social dominance. As the scale assesses a temperamental dimension, i.e. a personality trait rather than an acute episode, subjects are instructed to respond how they feel in general. Several studies have demonstrated the scale’s reliability and validity. The HPS is stable over time 16,17 , has been associated with psychiatric risk genes 18,19 , aggregates in families of afflict- ed patients 20 and has discriminated bipolar patients from controls 21,22 . Cross-sectionally, HPS high scorers showed higher rates of (hypo)manic and depressed episodes and more psychosocial im- pairment and substance use 16,23,24 . Longitudinally, the HPS predicted BD and hypomanic symptoms in addition to related disorders such as substance abuse 25–27 . However, it has recently been suggested that the HPS should be separated into different subscales which may have divergent psychopathological correlates 28-30 .
Sleep in Hypomanic Personality To date, only two small studies using student samples have assessed the association between sleep and HPS. In HPS high scorers, greater intraindividual variability in sleep duration was observed via sleep diaries 31 and actigraphy 32 . The latter study 32 also showed shorter sleep duration and greater intraindividual variability in sleep efficiency in HPS high scorers. As is usually the case, these two
38
studies utilized the HPS total sum score. In two recent studies 28,30 , HPS subscales were derived from factor analyses which differed in their associations with psychopathologically relevant traits 28,30,33 .
Methods Subjects The study sample was drawn from the LIFE-Adult study 34 , a population-based cohort comprised of 10,000 inhabitants of the city of Leipzig, Germany. Of the total sample, participants aged 60–82 years completed the HPS. Subjects had to be free of diseases or medications which could strongly impact sleep–wake behavior. Subjects were excluded when at current use of CNS-affecting drugs. Based on data from structured clinical interviews for DSM-IV Axis I disorders, we selected subjects without a lifetime history of substance dependence, psychotic or BDs, and who were free of current affective or anxiety disorders. In addition, participants were required to have available data from either the Pittsburgh sleep quality index (PSQI) assessment or an actigraphy recording for at least five nights. This resulted in a final sample of 771 subjects for actigraphy association analyses (372 female, Mage = 70.3 y), and 1766 subjects for PSQI association analyses (835 female, M age = 69.6 y). Factor analyses of the HPS were conducted using all 2861 subjects with complete HPS responses (1371 female; age range: 60–82 years; M age = 70.0).
Sleep Data To obtain objective measurements of sleep, subjects wore the SenseWear Pro 3 Armband acti- graph for an average of 6.9 days (range: 5–7 days). Subjective ratings of sleep and sleep quality were obtained using the German version of the PSQI 35 , a self-rating instrument to assess sleep quality during the past 4 weeks.
The Hypomanic Personality Scale The HPS, a self-rating scale used to assess BD risk25 , was developed with young subjects, which is reflected in some of the 48 items. As the current study applied the HPS to elderly subjects, four items from the German translation 36 were deleted for reasons of compliance. Associations between the sleep variables and the HPS were conducted for both the HPS total sum score (HPS total) and factor scores for each HPS subscale. We conducted partial Spearman corre- lations adjusting for sex and age.
Results: The factor analysis of the HPS revealed three factors, which we labeled “hypomanic core”, “social vitality” and “ordinariness”. Spearman correlations (Table 1) showed, that HPS total scale was cor- related with significantly shorter sleep duration, a greater number of awakenings, more time after sleep-onset (WASO) and lower sleep efficiency in HPS total. HPS total correlated even stronger with 39
night-to-night variability of the sleep parameters. On subscale level, sleep-wake variables were also significantly associated in the same direction with hypomanic core. In subjective sleep data (see PSQI data in Table 1), HPS total correlated with more daytime sleepiness and lower sleep quality, which was even stronger on subscale level in hypomanic core, which was also associated with longer latency of sleep-onset, short sleep duration, lower sleep efficiency and more daytime sleepiness. Social vitality and ordinariness were associated with less impaired subjective sleep pa- rameters.
Table 1 Partial Spearman correlations between hypomanic personality and sleep-wake variables
HPS total HPS subscale HPS subscale HPS subscale
sum-score Hypomanic Core Social Vitality Ordinariness rho p rho p rho p rho p
Actigraphy (n = 771) Means Sleep-onset latency .038 .294 .042 .240 .054 .134 -.035 .329 Sleep-onset time -.006 .864 -.035 .329 .025 .497 .061 .089 Sleep-offset time -.013 .721 -.041 .252 .007 .851 .056 .118 Sleep duration -.079 .029* -.072 .046* -.059 .101 .034 .350 NWAK .078 .030* .069 .056 .069 .056 -.011 .757 WASO .103 .004** .091 .011* .080 .027* -.046 .203 Sleep efficiency -.106 .003** -.101 .005** -.086 .017* .033 .361 Night-to-night variability Sleep-onset latency .019 .594 -.005 .893 .076 .036* -.026 .470 Sleep-onset time .122 7E-4** .102 .004** .113 .002** .009 .813 Sleep-offset time .110 .002** .119 9E-4** .069 .057 -.062 .087 Sleep duration .098 .006** .092 .011* .097 .007** -.003 .931 NWAK .145 6E-5** .109 .002** .152 2E-5** -.092 .011** WASO .115 .001** .116 .001** .095 .008** -.109 .002** Sleep efficiency .106 .003** .095 .008** .103 .004** -.092 .011** PSQI (n = 1766) Sleep-onset latency a .013 .576 .078 .001** -.080 8E-4** -.098 4E-5** Bedtime a,b .029 .230 .018 .444 .041 .087 .001 .968 Get-up time a -.022 .348 -.023 .344 -.020 .413 -.013 .581 Sleep duration a -.038 .114 -.095 7E-5** .027 .261 .081 6E-4** Sleep efficiency c .002 .925 -.065 .006** .071 .003** .091 1E-4** Daytime sleepiness a .075 .002** .088 2E-4** .052 .028* -.086 3E-4** PSQI score d .063 .008** .158 3E-11** -.063 .008** -.165 3E-12** Night-to-night variability is operationalized by intraindividual standard deviation (ISD) across a single subject's multiple nights. Note that Hypomanic Core, Social Vitality and Ordinariness here refer to factor scores derived from factor analyses as described in the Methods of the Supplementum. Results were additionally confirmed by analyses with traditional sum scores (see section 3.3). Effects of sex and age were partialled out. * p < .05, ** p < .01; NWAK : Number of awakenings; WASO : wake after sleep-onset time; a based on the respective Pittsburgh Sleep Quality Index (PSQI) item b time subject goes to bed; c quotient of sleep duration and time in bed with the latter calculated from bedtime and get-up time; d PSQI total score calculated according the manual from all PSQI
components; higher PSQI scores mean worse sleep quality.
40
Discussion: The current study analyzed whether increased vulnerability to BD, as assessed by the HPS, is linked to more disturbed sleep in healthy subjects. Correlation analyses were conducted between objective as well as subjective sleep parameters and HPS subscales and total scale. The correlation analyses as well as the extreme group comparisons revealed that a higher HPS total score is associated with worse sleep, greater night-to-night sleep variability and more daytime sleepiness. Thus, results confirm findings of impaired sleep in genetic high risk studies 12 . A noteworthy finding is that the HPS was more strongly associated with the night-to-night variability than with the mean sleep variables. In line with this, irregularity in sleep/wake behavior has been reported for euthymic and manic patients 2,37 and subjects with genetically or psychometrically operationalized heightened BD risk2,12,31,32,38,39 . Another finding is, that the HPS subscale hypomanic core correlations with impaired sleep variables were significantly higher than of HPS total. In contrast, the subscales ordinariness and social vitality showed associations with better perceived sleep. Thus, the current study strongly supports the suggestion of utilizing HPS subscales 28-30, 33 . The high scored daytime sleepiness given in subjective data and long sleep-onset latencies may occur due to being overly excited and reluctanct to fall asleep. The associations with shorter sleep are in line with studies conducted on young HPS high scorers and children at genetic risk for BD 32,40 . This stands in contrast to findings of longer sleep in euthymic BD1,2 . This might imply that long sleep may not be a preexisting factor, but instead part of the disease or a consequence of sedating medications, unemployement or psychotherapeutical interventions for longer sleep to avoid mania.
Strenghts and limitations: Our subscales’ correlations with sleep variables need replication in other, particularly younger samples. We showed that the HPS is also applicable in subjects older than 60 years. The older age of our sample supports our aim to associate the vulnerability factor HPS with sleep, as our subjects are unlikely to develop BD in the future. The present study was well-powered and revealed strong evidence for a link between hypomanic temperament and sleep alterations even though the effect sizes were small in correlation analyses.
Conclusion: The assiciations of the HPS with worse sleep and higher inter-night variability were consistent and especially HPS subscales support the suggestion of the HPS as a diagnostic and vulnerability factor. It supports the suggestion that impaired sleep may be a predisposing factor for BD. Interventions in improving sleep in HPS high scorers may be a valuable, modifiable and easy to access early prevention approach, which is largely free of social stigmatization.
41
IV) ReferenceReferences:s:s:s: Introduction: 1. Alonso J, Petukhova M, Vilagut G, et al. Days out of role due to common physical and mental conditions: results from the WHO World Mental Health surveys. Mol Psychiatry. 16, 1234–1246 (2011). 2. Grande, I., Berk, M., Birmaher, B., & Vieta, E. Bipolar disorder. The Lancet, 387(10027), 1561–1572 (2016). 3. Goodwin F, Jamison K. Manic-depressive illness: bipolar disorders and recurrent depression, 2nd edn. New York: Oxford University Press (2007). 4. Geoffroy PA, Scott J, Boudebesse C, et al. Sleep in patients with remitted bipolar disorders: a meta-analysis of actigraphy studies. Acta Psychiatr Scand.131(2), 89-99 (2015.) 5. Ng TH, Chung KF, Ho FY, Yeung WF, Yung KP, Lam TH. Sleep-wake disturbance in interepisode bipolar disorder and high-risk individuals: a systematic review and meta-analysis. Sleep Med Rev. 20, 46-58 (2015). 6. Barbini B, Bertelli S, Colombo C, Smeraldi E. Sleep loss, a possible factor in augmenting manic episode. Psychiatry Res. 65(2), 121- 125 (1996). 7. Bauer M, Grof P, Rasgon N, Bschor T, Glenn T, Whybrow PC. Temporal relation between sleep and mood in patients with bipolar disorder. Bipolar Disorders. 8(2),160-167 (2006). 8. Gruber J, Miklowitz DJ, Harvey AG, et al. Sleep matters: sleep functioning and course of illness in bipolar disorder. J Affect Disord. 134(1-3), 416-420 (2011). 9. Lewis KS, Gordon-Smith K, Forty L, et al. Sleep loss as a trigger of mood episodes in bipolar disorder: individual differences based on diagnostic subtype and gender. Br J Psychiatry (2017). 10. Sylvia LG, Dupuy JM, Ostacher MJ, et al. Sleep disturbance in euthymic bipolar patients. J Psychopharmacol. 26(8),1108-1112 (2012). 11. Wehr TA. Sleep loss: a preventable cause of mania and other excited states. J Clin Psychiatry. 50 Suppl:8-16, discussion 45-117 . (1989). 12. Wehr TA, Sack DA, Rosenthal NE. Sleep reduction as a final common pathway in the genesis of mania. Am J Psychiatry. 144(2), 201-204 (1987). 13. Harvey AG, Soehner AM, Kaplan KA, et al. Treating insomnia improves mood state, sleep, and functioning in bipolar disorder: a pilot randomized controlled trial. J Consult Clin Psychol. 83(3), 564-577 (2015). 14. Barbini B, Benedetti F, Colombo C, et al. Dark therapy for mania: a pilot study. Bipolar Disord.7(1), 98-101 (2005). 15. Plante DT, Winkelman JW. Sleep disturbance in bipolar disorder: therapeutic implications. Am J Psychiatry. 165(7), 830-843 (2008). 16. Kaplan KA, McGlinchey EL, Soehner A, et al. Hypersomnia subtypes, sleep and relapse in bipolar disorder. Psychol Med. 45(8),1751-1763 (2015). 17. Jawinski P, Kittel J, Sander C, et al. Recorded and Reported Sleepiness: The Association Between Brain Arousal in Resting State and Subjective Daytime Sleepiness. Sleep. 40(7) (2017). 18. Hegerl U, Stein M, Mulert C, Mergl R, Olbrich S, Dichgans E, Rujescu D, Pogarell O. EEG-vigilance differences between patients with borderline personality disorder, patients with obsessive-compulsive disorder and healthy controls. Eur Arch Psychiatry Clin Neurosci.Apr. 258(3), 137-143 (2008). 19. Olbrich S, Mulert C, Karch S, Trenner M, Leicht G, Pogarell O, Hegerl U. EEG-vigilance and BOLD effect during simultaneous EEG/fMRI measurement. Neuroimage.Apr 1. 45(2), 319-332 (2009). 20. Olbrich S, Sander C, Minkwitz J, Chittka T, Mergl R, Hegerl U, Himmerich H. EEG vigilance regulation patterns and their discriminative power to separate patients with major depression from healthy controls. Neuropsychobiology.Jun. 65(4),188-194 (2012). 21. Hegerl U, Hensch T. The vigilance regulation model of affective disorders and ADHD. Neurosci Biobehav Rev. 44, 45-57 (2014). 22. Hegerl U, Himmerich H, Engmann B, Hensch T. Mania and attention-deficit/hyperactivity disorder: common symptomatology, common pathophysiology and common treatment? Curr Opin Psychiatry. 23(1),1-7 (2010). 23. Hegerl U, Sander C, Hensch T. Arousal Regulation in Affective Disorders. In: Frodl T, ed. Systems Neuroscience in Depression. Amsterdam: Academic Press. 341-370 (2016). 24. Wittekind DA, Spada J, Gross A, et al. Early report on brain arousal regulation in manic vs depressive episodes in bipolar disorder. Bipolar Disord. 18(6), 502-510 (2016). 25. Maoz H, Goldstein T, Axelson DA, et al. Dimensional psychopathology in preschool offspring of parents with bipolar disorder. J Child Psychol Psychiatry. 55(2),144-153 (2014). 26. Melo MC, Garcia RF, Linhares Neto VB, et al. Sleep and circadian alterations in people at risk for bipolar disorder: A systematic review. J Psychiatr Res. 83, 211-219 (2016). 27. Pearson, J. S., & Kley, I. B. On the application of genetic expectancies as age-specific base rates in the study of human behavior disorders. Psychological Bulletin. 54(5), 406-420 (1957). 28. Chapman LJ, Chapman JP, Kwapil TR, Eckblad M, Zinser MC. Putatively psychosis-prone subjects 10 years later. Journal of Abnormal Psychology. 103, 171–183 (1994). 29. Cicero DC, Martin EA, Becker TM, Docherty AR, Kerns JG. Correspondence between psychometric and clinical high risk for psychosis in an undergraduate population. Psychol Assess. 26(3), 901-915 (2014). 30. Fusar-Poli, P., Cappucciati, M., Rutigliano, G., Schultze-Lutter, F., Bonoldi, I., Borgwardt, S., Riecher-Rössler, A., Addington, J., Perkins, D., Woods, S. W., McGlashan, T. H., Lee, J., Klosterkötter, J., Yung, A. R., McGuire, P. At risk or not at risk? A meta-analysis of the prognostic accuracy of psychometric interviews for psychosis prediction. World psychiatry : official journal of the World Psychiatric Association (WPA). 14(3), 322-332 (2015). 31. Reiss E., Konstitutionelle Verstimmung und manisch-depressives Irresein. Z Ges Neurol Psychiatrie 2, 347-628 (1910). 32. Kraepelin E., Psychiatrie. Ein Lehrbuch für Studierende und Ärzte, vol III. (8th edn). Barth. Leipzig (1913). 33. Tellenbach H., Melancholie. 1st edn, Springer, Berlin Heidelberg New York (1961). 34. Tellenbach H. Melancholy (English trans. Of 3rd German edn). Dusquesne University Press, Pittsburgh (1980). 35. Ambrosini A, Stanghellini G, Langer AI. Typus melancholicus from tellenbach up to the present day: a review about the premorbid personality vulnerable to melancholia. Actas Esp Psiquiatr. 39(5): 302-311 (2011). 36. Stanghellini G, Mundt C. Personality and endogenous/major depression: an empirical approach to typus melancholicus. Theoretical Issues. Psychopathology . 30, 119-129 (1997). 37. Lauer CJ, von Zerssen D, Schreiber W, Modell S, Holsboer F, Krieg JC. The pre-morbid psychometric profile is stable over time in subjects at high familial risk for affective disorders. J Affect Disord. 51, 45-53 (1998). 38. Hecht H, van Calker D, Berger M, von Zerssen D. Personality in patients with affective disorders and their relatives. J Affect Disord. 51, 33-43 (1988). 39. von Zerssen D. Premorbid personality and affective psychoses in Burrows DG (ed): Handbook of Studies on Depression. Section I: 42
Studies in classification, phenomenology and aetiology of depression. Amsterdam, Excerpta Medica. (1977) 40. Akiskal, H.S. The temperamental foundations of affective disorders. In C. Mundt, M.J. Goldstein, K. Hahlweg & P. Fiedler (Eds), Interpersonal factors in the origin and course of affective disorder (pp. 3-30) (1996). 41. Eckblad M, Chapman, LJ. Development and validation of a scale for hypomanic personality. Journal of Abnormal Psychology. 95, 214– 222 (1986). 42. Meyer, T. D. The Hypomanic Personality Scale, the Big Five, and their relationship to depression and mania: Personality and Individual Differences. 32(4), 649-660 (2002). 43. Meyer TD, Hautzinger M. Screening for bipolar disorders using the Hypomanic Personality Scale. J Affect Disord. 75(2),149-154 (2003). 44. Blechert J, Meyer TD., Are measures of hypomanic personality, impulsive nonconformity and rigidity predictors of bipolar symptoms? Br J Clin Psychol. 44(1), 15-27 (2005.). 45. Walsh MA, DeGeorge DP, Barrantes-Vidal N, Kwapil TR. A 3-Year Longitudinal Study of Risk for Bipolar Spectrum Psychopathology. J Abnorm Psychol. 124(3), 486-497 (2015). 46. Kwapil TR, Raulin ML, Midthun JC. A ten-year longitudinal study of intense ambivalence as a predictor of risk for psychopathology. J Nerv Ment Dis. 188(7):402-408 (2000). 47. Burmeister M, McInnis MG, Zollner S. Psychiatric genetics: progress amid controversy. Nat Rev Genet. 9(7), 527-540 (2008). 48. Savitz JB, Ramesar RS. Personality: is it a viable endophenotype for genetic studies of bipolar affective disorder? BipolarDisord. 8(4), 322-337 (2006). 49. Cuthbert BN. Research Domain Criteria: toward future psychiatric nosologies. Dialogues Clin Neurosci.17(1), 89-97 (2015). 50. Meyer TD, Maier S. Is there evidence for social rhythm instability in people at risk for affective disorders? Psychiatry Research. 141(1),103-114 (2006). 51. Ankers D, Jones SH. Objective assessment of circadian activity and sleep patterns in individuals at behavioural risk of hypomania. J Clin Psychol. 65(10),1071-1086 (2009).
Summary: 1. Geoffroy, P. A. et al. Sleep in patients with remitted bipolar disorders: a metaanalysis of actigraphy studies. Acta Psychiatr. Scand. 131, 89–99 (2015). 2. Ng, T. H. et al. Sleep-wake disturbance in interepisode bipolar disorder and high-risk individuals: a systematic review and meta- analysis. Sleep. Med. Rev. 20, 46–58 (2015). 3. Barbini, B., Bertelli, S., Colombo, C. & Smeraldi, E. Sleep loss, a possible factor in augmenting manic episode. Psychiatry Res. 65, 121–125 (1996). 4. Bauer, M. et al. Temporal relation between sleep and mood in patients with bipolar disorder. Bipolar Disord. 8, 160–167 (2006). 5. Gruber, J. et al. Sleep matters: sleep functioning and course of illness in bipolar disorder. J. Affect. Disord. 134, 416–420 (2011). 6. Lewis, K. S. et al. Sleep loss as a trigger of mood episodes in bipolar disorder: individual differences based on diagnostic subtype and gender. Br. J. Psychiatry 211, 169–174 (2017). 7. Sylvia, L. G. et al. Sleep disturbance in euthymic bipolar patients. J. Psychopharmacol. 26, 1108–1112 (2012). 8. Wehr, T. A. Sleep loss: a preventable cause of mania and other excited states. J. Clin. Psychiatry 50(Suppl), 8–16 (1989). Discussion 45-17. 9. Wehr, T. A., Sack, D. A. & Rosenthal, N. E. Sleep reduction as a final common pathway in the genesis of mania. Am. J. Psychiatry 144, 201–204 (1987). 10. Kaplan, K. A. et al. Hypersomnia subtypes, sleep and relapse in bipolar disorder. Psychol. Med. 45, 1751–1763 (2015). 11. Maoz, H. et al. Dimensional psychopathology in preschool offspring of parents with bipolar disorder. J. Child Psychol. Psychiatry 55, 144–153 (2014). 12. Melo, M. C. et al. Sleep and circadian alterations in people at risk for bipolar disorder: a systematic review. J. Psychiatr. Res. 83, 211–219 (2016). 13. Pancheri, C. et al. A systematic review on sleep alterations anticipating the onset of bipolar disorder. Eur. Psychiatry 58, 45–53 (2019). 14. Ritter, P. S. et al. Disturbed sleep as risk factor for the subsequent onset of bipolar disorder-Data from a 10-year prospective- longitudinal study among adolescents and young adults. J. Psychiatr. Res. 68, 76–82 (2015). 15. Miller, G. A. The Behavioral High-Risk Paradigm in Psychopathology. (Springer, 1995). 16. Eckblad, M. & Chapman, L. J. Development and validation of a scale for hypomanic personality. J. Abnorm. Psychol. 95, 214–222 (1986). 17. Hofmann, B. U. &Meyer, T. D.Mood fluctuations in people putatively at risk for bipolar disorders. Br. J. Clin. Psychol. 45, 105–110 (2006). 18. Johnson, S. L., Carver, C. S., Joormann, J. & Cuccaro, M. A genetic analysis of the validity of the Hypomanic Personality Scale. Bipolar Disord. 17, 331–339 (2015). 19. Ortega-Alonso, A. et al. Genome-wide association study of psychosis proneness in the Finnish population. Schizophrenia Bull. 43, 1304–1314 (2017). 20. Meyer, T. D. & Hautzinger, M. Hypomanic personality, social anhedonia and impulsive nonconformity: evidence for familial aggregation? J. Personal. Disord. 15, 281–299 (2001). 21. Miller, C. J., Johnson, S. L., Kwapil, T. R. & Carver, C. S. Three studies on selfreport scales to detect bipolar disorder. J. Affect. Disord. 128, 199–210 (2011). 22. Rowland, J. E. et al. Cognitive regulation of negative affect in schizophrenia and bipolar disorder. Psychiatry Res. 208, 21–28 (2013). 23. Klein, D. N., Lewinsohn, P. M. & Seeley, J. R. Hypomanic personality traits in a community sample of adolescents. J. Affect. Disord. 38, 135–143 (1996). 24. Meyer, T. D. & Hautzinger, M. Screening for bipolar disorders using the Hypomanic Personality Scale. J. Affect. Disord. 75, 149–154 (2003). 25. Kwapil, T. R. et al. A longitudinal study of high scorers on the hypomanic personality scale. J. Abnorm. Psychol. 109, 222–226 (2000). 26. Walsh, M. A., DeGeorge, D. P., Barrantes-Vidal, N. & Kwapil, T. R. A 3-year longitudinal study of risk for bipolar spectrum psychopathology. J. Abnorm. Psychol. 124, 486–497 (2015). 27. Blechert, J. & Meyer, T. D. Are measures of hypomanic personality, impulsive nonconformity and rigidity predictors of bipolar symptoms? Br. J. Clin. Psychol. 44, 15–27 (2005). 28. Schalet, B. D., Durbin, C. E. & Revelle, W. Multidimensional structure of the Hypomanic Personality Scale. Psychol. Assess. 23, 504– 43
522 (2011). 29. Stanton, K., Gruber, J. & Watson, D. Basic dimensions defining mania risk: a structural approach. Psychol. Assess. 29, 304–319 (2017). 30. Terrien, S., Stefaniak, N., Morvan, Y. & Besche-Richard, C. Factor structure of the French version of the Hypomanic Personality Scale (HPS) in non-clinical young adults. Compr. Psychiatry 62, 105–113 (2015). 31. Meyer, T. D. & Maier, S. Is there evidence for social rhythm instability in people at risk for affective disorders? Psychiatry Res. 141, 103–114 (2006). 32. Ankers, D. & Jones, S. H. Objective assessment of circadian activity and sleep patterns in individuals at behavioural risk of hypomania. J. Clin. Psychol. 65, 1071–1086 (2009). 33. Ford, B. Q., Mauss, I. B. & Gruber, J. Valuing happiness is associated with bipolar disorder. Emotion 15, 211–222 (2015). 34. Loeffler, M. et al. The LIFE-Adult-Study: objectives and design of a populationbased cohort study with 10,000 deeply phenotyped adults in Germany. BMC Public Health 15, 691 (2015). 35. Hinz, A. et al. Sleep quality in the general population: psychometric properties of the Pittsburgh Sleep Quality Index, derived from a German community sample of 9284 people. Sleep Med. 30, 57–63 (2017). 36. Meyer, T. D., Drüke, B. & Hautzinger, M. Hypomane Persönlichkeit-Psychometrische Evaluation und erste Ergebnisse zur Validität der deutschen Version der Chapman-Skala. Z. Klin. Psychol. Psychother. 29, 35–42 (2000). 37. Scott, J. et al. Activation in bipolar disorders: a systematic review. JAMA Psychiatry 74, 189–196 (2017). 38. Duffy, A., Jones S., Goodday S. & Bentall R. Candidate risks indicators for bipolar disorder: early intervention opportunities in high- risk youth. Int. J. Neuropsychopharmacol. 19, pyv071, https://doi.org/10.1093/ijnp/pyv071 (2015). 39. Hoaki, N. et al. Biological aspect of hyperthymic temperament: light, sleep, and serotonin. Psychopharmacology 213, 633–638 (2011). 40. Egeland, J. A. et al. A 16-year prospective study of prodromal features prior to BPI onset in well Amish children. J. Affect. Disord. 142, 186–192 (2012).
44
VIVIVI)VI ) Declaration of AuthorsAuthorshiphiphiphip::::
Hiermit erkläre ich, dass ich die vorliegende Arbeit selbstständig und ohne unzulässige Hilfe oder Benutzung anderer als der angegebenen Hilfsmittel angefertigt habe. Ich versichere, dass Dritte von mir weder unmittelbar noch mittelbar eine Vergütung oder geldwerte Leistungen für Arbeiten erhalten haben, die im Zusammenhang mit dem Inhalt der vorgelegten Dissertation stehen, und dass die vorgelegte Arbeit weder im Inland noch im Ausland in gleicher oder ähnlicher Form einer anderen Prüfungsbehörde zum Zweck einer Promotion oder eines anderen Prüfungsverfahrens vorgelegt wurde. Alles aus anderen Quellen und von anderen Personen übernommene Material, das in der Arbeit verwendet wurde oder auf das direkt Bezug genommen wird, wurde als solches kenntlich gemacht. Insbesondere wurden alle Personen genannt, die direkt an der Entstehung der vorliegenden Arbeit beteiligt waren. Die aktuellen gesetzlichen Vorgaben in Bezug auf die Zulassung der klinischen Studien, die Bestimmungen des Tierschutzgesetzes, die Bestimmungen des Gentechnikgesetzes und die allgemeinen Datenschutzbestimmungen wurden eingehalten. Ich versichere, dass ich die Regelungen der Satzung der Universität Leipzig zur Sicherung guter wissenschaftlicher Praxis kenne und eingehalten habe.
...... Datum David Wozniak
46
VIVIVII)VI I) Curriculum VitaeVitae::::
David Wozniak Geboren am 29.07.1990 in Berlin Dresdner Straße 82, 04317 Leipzig [email protected]
Akademische Ausbildung: 2010: Abitur am Anton Bruckner Gymnasium in Straubing (Note 1,4) 2011-2017: Studium der Humanmedizin an der Universität Leipzig (Ärztliche Approbation) 17.09.2013: 1. Staatsexamen 13.10.2016: 2. Staatsexamen 18.12.2017: Ärztliche Approbation
Aktuelle Tätigkeit: Seit 2018: Arzt in Weiterbildung in der Klinik für Psychiatrie und Psychotherapie des Universitätsklinikums Leipzig
Zusätzliche Tätigkeiten: Seit 2018: Psychotherapieausbildung kognitive Verhaltenstherapie Seit 2018: Prüfarzt Seit 2018: Referent am Medizindidaktischem Zentrum Leipzig und PJ-Betreuung Seit 2019: Studienarzt in der Vagusnervstudie bei PD Dr. med. Frank Schmidt
Schaffen als freischaffender Autor und Schauspieler, u.a. 2015: Regie und Drehbuch Kurzfilm „Die Füchsin 2016: Kurzgeschichtenband „Juleika Lippenrot“, chili Verlag 2017: Hörspiel „Warum bist du nicht Sussja gewesen?“ 2019: Rolle im Kinofilm „Das Melancholische Mädchen“ (engl. „Aren’t You Happy?“)
47
VIIIVIII)) PublPublicationsicationsications::::
Wozniak D., Dietzel J. Fallbericht: Late-Onset Huntington’s Disease bei einer 81-jährigen Patientin ohne positive Familienanamnese. Poster präsentiert auf den Mitteldeutschen Psychiatrietagen. (2019)
Hensch, T.*, Wozniak, D.*, et al. Vulnerability to bipolar disorder is linked to sleep and sleepiness. Transl Psychiatry 9, 294. (2019)
*shared first authors
48
IX) AcknowledgementsAcknowledgements::::
All Diejenigen die mit Vorurteilen, Unwissen und Desinformation das schützende und helfende Wesen der empirischen Medizin bedrohen wenn ihr eine Danksagung einer Dissertation lest bin ich euch dankbar für einen Schritt in die richtige Richtung
All den Menschen und Dingen die mich um den Schlaf bringen im Guten, wie im Schlechten Danke!
49