Associations between sleep duration and suicidality in adolescents: A systematic review and dose–response meta-analysis

Associations between sleep duration and suicidality in adolescents: A systematic review and dose–response meta-analysis

Accepted Manuscript Associations Between Sleep Duration and Suicidality in Adolescents: A Systematic Review and Dose–Response Meta-Analysis Hsiao-Yean...

358KB Sizes 0 Downloads 100 Views

Accepted Manuscript Associations Between Sleep Duration and Suicidality in Adolescents: A Systematic Review and Dose–Response Meta-Analysis Hsiao-Yean Chiu, Hsin-Chien Lee, Pin-Yuan Chen, Ying-Fan Lai, Yu-Kang Tu PII:

S1087-0792(17)30167-3

DOI:

10.1016/j.smrv.2018.07.003

Reference:

YSMRV 1116

To appear in:

Sleep Medicine Reviews

Received Date: 28 August 2017 Revised Date:

4 July 2018

Accepted Date: 5 July 2018

Please cite this article as: Chiu H-Y, Lee H-C, Chen P-Y, Lai Y-F, Tu Y-K, Associations Between Sleep Duration and Suicidality in Adolescents: A Systematic Review and Dose–Response Meta-Analysis, Sleep Medicine Reviews (2018), doi: 10.1016/j.smrv.2018.07.003. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Associations Between Sleep Duration and Suicidality in Adolescents: A Systematic Review and Dose–Response Meta-Analysis Running Heading: Sleep Duration and Suicidality Hsiao-Yean Chiu a,b, Hsin-Chien Lee b,c, Pin-Yuan Chen d,e, Ying-Fan Lai a, Yu-Kang Tu f

RI PT

AFFILIATIONS: a

School of Nursing, College of Nursing, Taipei Medical University, Taipei, Taiwan.

b

Research Center of Sleep Medicine, School of Medicine, Taipei Medical University, Taipei,

c

SC

Taiwan.

Department of Psychiatry, Taipei Medical University Shuang Ho Hospital, New Taipei City,

M AN U

Taiwan. d

Department of Neurosurgery, Chang Gung Memorial Hospital, Keelung, Taiwan

e

School of Medicine, Chang Gung University, Taoyuan, Taiwan.

f

Department of Public Health, Institute of Epidemiology and Preventive Medicine, College of

Corresponding author Yu-Kang Tu

TE D

Public Health, National Taiwan University, Taipei, Taiwan.

EP

Address: 17, Xu-Zhou Road, Taipei, Taiwan E-mail: [email protected]

AC C

Tel.: +886 2 33668039 Fax: +886 2 23511955 Acknowledgment

The authors declare no potential conflicts of interest regarding the authorship and publication of this article. This meta-analysis was supported by grants from the Ministry of Science and Technology, Taiwan (MOST 106-2314-B-038 -058 -MY3 and MOST 106-2314-B-002 -098 -MY3).

1

ACCEPTED MANUSCRIPT Summary

Sleep duration has received considerable attention as a potential risk factor of suicidality in youths; however, evidence on the dose–response association between sleep duration and

RI PT

suicidality has not been synthesized. This meta-analysis examined linear and nonlinear dose– response relationships between sleep duration and the risk of suicidality in adolescents and explored potential moderators of the associations. Electronic databases, namely EMBASE,

SC

PubMed, PsycINFO, Wanfang Data (Chinese database), and the China Knowledge Resource

M AN U

Integrated Database, were searched from their inception to April 18, 2017. Studies examining the association between sleep duration and suicidality in adolescents were included. A random-effects dose–response model was used to estimate the linear and nonlinear dose–

TE D

response relationships. We identified 13 reports that included a total of 598,281 participants for a systematic review, and 12 reports were further used for a dose–response meta-analysis. Strong curvilinear dose–response associations were obtained for both suicidal ideation and

EP

attempts, with the lowest suicidal ideation and attempt risks at sleep durations of 8 h and 8–9

AC C

h per day (all Pnonlinearity < .001). A linear dose–response relationship between sleep duration and suicide plans (pooled OR = 0.89, 95% confidence interval [CI] = 0.88–0.90) was obtained, indicating that the risk of suicide plans statistically decreased by 11% for every 1-h increase in sleep duration. Depression did not moderate the association between sleep duration and suicidality in youths. Our findings suggest curvilinear dose–response associations between sleep duration and the risks of suicidal ideation and attempts and a linear dose–response 2

ACCEPTED MANUSCRIPT relationship between sleep duration and suicide plan risk. Additional longitudinal studies are warranted to establish causality.

AC C

EP

TE D

M AN U

SC

RI PT

Keywords: Dose–response meta-analysis, sleep duration, suicidality, adolescents

3

ACCEPTED MANUSCRIPT INTRODUCTION Suicide in youth is a major public health concern. It has a global rate of 7.4 per 100 000 persons [1] and thus is the third leading cause of death in young boys and the second leading

RI PT

cause of death in young girls [2]. Suicidality is defined as suicidal ideation, plans, attempts, and suicide itself [3, 4]. The estimated worldwide lifetime prevalence for suicidal ideation

SC

ranged from 12.1% to 31.5% and that for suicide attempts ranged from 4.1% to 23.5% [5–8]. Identifying potentially modifiable risk factors of suicidality in adolescents is clinically

M AN U

relevant for reducing youth suicide rates.

Although several risk factors for youth suicidality have been identified, namely the female sex [9, 10], psychiatric or mental disorders [11], exposure to bullying and violence [12,

TE D

13], substance abuse [14], weak family functioning [15]and low acceptance from peer groups [16], sleep duration has recently received considerable attention. The National Sleep

EP

Foundation recommends that the appropriate sleep duration for adolescents is between 8 and

AC C

10 h [17]; however, because of social factors (eg, academic requirements or electronic media) [18] and biological characteristics (e.g., the tendency towards delayed sleep phase coupled with fixed wake up times) (19, 20), more than half of adolescents (53.6%) sleep fewer than 8 h [21].

Over the past decade, several large-scale cross-sectional studies [22–34] have reported inconsistent findings on the association between sleep duration and the risk of youth suicidality. Some studies have suggested a negative association [23, 24]), whereas some have 4

ACCEPTED MANUSCRIPT reported that both shorter and longer sleep durations increased the risk of youth suicidality [25, 34]. Two previous meta-analyses that investigated the associations between sleep disturbances and suicidality in the general population [35] and in patients with psychiatric diagnoses [36]

RI PT

revealed that sleep disturbance (e.g., insomnia, nightmares, parasomnia, and sleep-related breathing disorder) substantially increased the risk of suicidality (relative risk = 1.95 to 2.95,

SC

and odds ratio [OR] = 1.99, respectively). At present, there is no systematic review and

meta-analysis that investigates the relevance of sleep duration to suicidality in adolescents.

M AN U

Because suicide is a major concern for adolescent health, the association between sleep duration and youth suicidality may be crucial for detecting, treating, and potentially preventing youth suicide. We therefore conducted a systematic review and dose–response

TE D

meta-analysis to investigate the association between sleep duration and the risks of suicidal ideation, plans, and attempts in adolescents. In addition, we examined the moderating effects

EP

of the participants’ characteristics and methodological features on the relationship between

AC C

sleep duration and suicidality. Methods

Search strategies and selection criteria This dose–response meta-analysis was conducted in accordance with preferred reporting items for systematic reviews and meta-analyses [37]. A systematic literature search was performed in all fields in EMBASE, PubMed, PsycINFO, ProQuest Dissertations & Theses A&I, Wanfang Data (Chinese database), and the China Knowledge Resource Integrated 5

ACCEPTED MANUSCRIPT Database from their inception to April 18, 2017. The combinations of keywords used are listed as follows: (sleep duration OR sleep time) AND (suicidal ideation OR suicidal plan OR suicidal attempt) AND (adolescent). An example of a search string is provided in Table S1.

RI PT

We also manually searched the reference lists of included studies and relevant published reports and contacted study authors.

SC

We included studies with a cross-sectional, case–control, or cohort study design that evaluated the association between sleep duration and suicidal ideation, plans, or attempts in

M AN U

adolescents. No language restriction was applied. If the data of the included studies were incomplete for the analyses, the authors would be contacted to obtain additional information. Two investigators (HYC and YFL) independently searched, screened, and evaluated all

TE D

relevant studies according to the eligibility criteria. Any disagreement was resolved through discussion with a third researcher (PYC).

EP

Data extraction and assessment of the study quality

AC C

Two investigators (HYC and YFL) independently extracted data from included studies by using a predesigned data extraction form, including the author names, publication year, country where the study was conducted, average age of the study population, percentage of female participants, sample size, outcome variables, and adjusted covariates. Disagreements were resolved through discussion. The Critical Appraisal Checklist recommended by the Joanna Briggs Institute was used to assess the methodological quality of the included studies [38]. The checklist comprises 8 6

ACCEPTED MANUSCRIPT items, and each item is rated as “yes,” “no,” or “unclear.” The included studies were independently evaluated by two researchers (HYC and YFL). Any disagreement was resolved by consensus through face-to-face discussion.

RI PT

Data synthesis and analysis The odds ratio (OR) with a 95% confidence interval (CI) was used as a measurement of

SC

the association between sleep duration and the risks of suicidal ideation, plans, and attempts. For the dose-response meta-analysis, the reported mean of each sleep duration category was

M AN U

used. When the ranges of sleep duration categories were reported, we used the average value of the lower and upper bounds of each category. When the lowest category was open ended, the average value of the upper bound and 0 was used. When the highest category was open

TE D

ended, the average value was assumed to be 1.2 times the lower boundary [39]. The highest versus the lowest category of sleep duration in the included studies was first

EP

analyzed using a random-effects meta-analysis (high versus low meta-analysis) [40]. In the

AC C

dose–response meta-analysis, only studies with at least 3 exposure categories were included (41). We first examined the linear trends of the association between exposure and outcomes by using the method described by Greenland and Longnecker [42, 43]. The estimated linear trends were then pooled using random-effects meta-analysis [40]. For the nonlinear dose– response trends, we used the restricted cubic splines with 4 knots (percentiles of 5%, 35%, 65%, and 95%) [42, 44], and the results from each study were then pooled using random-effects multivariate meta-analysis [45, 46]. A likelihood ratio test was performed to 7

ACCEPTED MANUSCRIPT compare the difference between the linear and nonlinear models [44]. When the lowest category was not the reference category,, we used the method described by Hamling et al [47]. to convert the reported odds ratios to those with the lowest category of sleep duration as the

RI PT

reference category. Heterogeneity between studies was assessed using the Q test and I2 statistics, which

SC

quantify the total variation explained by between-study variation. Subgroup analyses were performed to explain the observed heterogeneity among studies. In the current study, we did

M AN U

not give a score to assess study quality; however, we examined indicators of study quality, including sample size and the adjustment for depression, which might have modified our results. Additional factors such as age, the percentage of women, and the country where the

TE D

study was conducted were assessed using subgroup analyses and metaregression. Heterogeneity between subgroups was determined using metaregression. Furthermore, we

EP

performed a sensitivity analysis to examine the robustness of our findings by excluding 1

AC C

study at a time and recalculating the pooled estimates. RESULTS

Search results and study characteristics We initially identified 640 articles and screened their titles and abstracts (Figure 1). After duplicates and irrelevant articles were excluded, 10 cross-sectional studies [22-25, 27, 29-31, 34, 48] met the inclusion criteria and were thus included for our systematic review and meta-analysis. Because one study [34] reported rate ratios in two adolescent populations in 8

ACCEPTED MANUSCRIPT 2007 and 2009 and two [31, 48] exclusively reported sex-specific rate ratios, 13 reports were included for the final analyses. For the dose–response meta-analysis, we excluded one study [22] with only two sleep duration categories, because at least three exposure categories were

RI PT

required to estimate the study-specific trend. Twelve reports were then included for the dose– response meta-analysis. A list of studies excluded after a full-text review is presented in Table

SC

S2.

Overall, 598,281 adolescents with a mean age of 15.5 years (ranging from 14.6 to 16.3

M AN U

years) were included. The average percentage of female adolescents was 48.8%. Only one report was published in Chinese. Of the reports, seven were conducted in South Korea, three in China, and three in the United States. Sleep duration was measured using a self-reported

TE D

sleep duration frequency question in all of the included studies except one, which used the Pittsburgh Sleep Quality Index [22]. Of which, six reports [25, 30, 31, 48] focused on the

EP

weekday average sleep duration (Monday–Friday) and others investigated the mean sleep

AC C

duration on weekdays and weekends. Several reference categories of exposure were used in the included studies. Four reports used ≥8 h/d as the reference category, three used 8 h/d, two used 6–7 h/d, one used 7–8 h/d, one used 7 h/d, and one used ≥9 h/d. Most reports assessed suicidal ideation, plans, or attempts during the past 12 months using one question; one study [29] focused on the past 6 months. All studies adjusted for potential covariates (Table S3 and S4). Quality of Studies 9

ACCEPTED MANUSCRIPT The results of the Critical Appraisal Checklist assessment of the included studies are presented in Table S5. No article met all eight criteria for a high methodological quality. None of the studies applied the valid and reliable tools in measuring sleep duration and suicidality.

RI PT

High vs low metaanalysis In terms of the risk of suicidal ideation, 13 reports [22-25, 27, 29-31, 34, 48] were

SC

included in the high versus low meta-analysis. The pooled OR was 0.55 (95% CI = 0.47–0.65),

for those with greater sleep duration.

M AN U

with evidence of heterogeneity (P < .001, I2 = 92.5%, Figure S1A), suggesting a lower risk

With regards to the risk of suicidal plans, we included six reports [22, 27, 31, 34] in the analysis of high versus low meta-analysis (Figure S1B). The pooled OR was 0.50 (95% CI =

TE D

0.44–0.56) for those with the longest sleep duration compared with those with the shortest sleep duration, with no evidence of heterogeneity (P = .57, I2 = 0.0%).

EP

For the risk of youth suicide attempts, we included 11 reports (Figure S1C) [22, 25, 27,

AC C

29-31, 34, 48]. The pooled OR was 0.52 (95% CI = 0.41–0.66), with evidence of heterogeneity (P < .0001, I2 = 78.5%). Dose-response association between sleep duration and the risk of suicidal ideation, plans, and attempts

Figure 2A shows the results of the linear and nonlinear dose–response relationship between sleep duration and suicidal ideation. The likelihood ratio test suggested nonlinearity (P = .00003), indicating that there was a nonlinear relationship between sleep duration and 10

ACCEPTED MANUSCRIPT suicidal ideation (Pheterogeneity < .001, I2 = 95.4%), with a U-shaped dose–response curve. The lowest risk of suicidal ideation was observed with a sleep duration of 9 h. Figure 2B presents the results of the linear and nonlinear dose-response relationship

RI PT

between sleep duration and suicide plans. The likelihood ratio test suggested linearity (P = .15), and the pooled OR of suicide plans for a 1-h increase in sleep duration was 0.89 (95%

SC

CI = 0.88–0.90) (i.e., the risk of suicide plans decreased by 11% for every 1-h increase in sleep duration, with no heterogeneity across the included studies (P = .16, I2 = 39.7%).

M AN U

Figure 2C shows the results of the linear and nonlinear dose-response relationship between sleep duration and suicide attempts. The likelihood ratio test suggested nonlinearity (P = .0006), indicating a significant nonlinear dose–response relationship (P heterogeneity < .001,

TE D

I2 = 94.1%) with a U-shaped dose–response curve. The lowest risk of suicide attempts was observed with a sleep duration of 8–9 h.

EP

Subgroup analysis, metaregression, and sensitivity analyses

AC C

Subgroup analyses and metaregression analyses for the high versus low analyses are presented in Table 1. No significant factor explaining the observed heterogeneity was identified for suicidal ideation, plans, or attempts. Notably, the studies that adjusted for depression had an OR comparable with that of studies that did not. Sensitivity analyses demonstrated similar results (Table S6), indicating that the overall results were not significantly influenced by any single study, with ORs and 95% CIs ranging from 0.54 (0.46–0.63) to 0.57 (0.48–0.67) for suicidal ideation, 0.44 (0.36–0.55) to 0.51 11

ACCEPTED MANUSCRIPT (0.45–0.57) for suicide plans, and 0.49 (0.38–0.62) to 0.55 (0.43–0.70) for suicide attempts. DISCUSSION In this dose–response meta-analysis, we found a U-shaped association between sleep

RI PT

duration and youth suicidal ideation and attempts incidence. We also found a negative linear dose-response association in youth suicidal plans. Every 1-h increase in sleep duration led to a

factor for the development of youth suicidality.

SC

decreased risk of 11% of suicide plans in adolescents. Sleep duration is an independent risk

M AN U

Suicide is a multifactorial behavior, and our findings suggest that adequate sleep may be a protective factor that reduces the suicide rate in adolescents. Mechanisms linking sleep to suicidality may differ between short and long sleep durations. One possibility is that being

TE D

awake at night creates a window of vulnerability with respect to suicidality. That is, both sleep deprivation and circadian effects compromise daytime and nighttime frontal lobe

EP

function/executive function. Hypoactivation of the frontal lobe could diminish

AC C

problem-solving abilities and increase impulsive behavior, both of which may increase suicide risk [49]. The serotonergic system has been proposed to mediate the association between short sleep duration and suicide [50]. A previous study reported that the prefrontal cortex exhibited low serotonin synthesis in suicide attempters compared with healthy controls [51]. Lower neuron density and deficient serotonin input in the prefrontal cortex, which controls executive function, may contribute to impulsive aggressive traits that cause suicidality [52, 53]. Serotonin and its receptors in the brain also play a critical role in sleep– 12

ACCEPTED MANUSCRIPT wake regulation. Serotonin secretion is highest during wakefulness and decreases during sleep [54]. Increasing evidence suggests that restricted sleep results in loss of sensitivity or desensitization of postsynaptic serotonin receptors (55, 56). Under these scenarios, reduced

RI PT

sleep duration might lead to a loss of serotonin function and thus adversely affect impulse control and judgment, which in turn increases the likelihood of suicide.

SC

Explanations for the relationship between long sleep duration and increased suicidality are currently speculative. The relevant literature demonstrates that depressive symptoms play

M AN U

a moderating role in the linkage between a long sleep duration and suicidality, as evidenced by the relationship disappearing after adjustment for depressive mood [57]. Nonetheless, our results from subgroup analyses indicated no moderating role, which is consistent with

TE D

previous meta-analyses [35, 36], showing that depression did not moderate the association between sleep disturbances and suicidality. Collectively, studies have suggested that the

EP

maintenance of sleep may be involved in suicidality independent of depression. The elevation

AC C

of serotonergic activity caused by increased sleep duration might be a possible explanation [50]. Because the underlying mechanisms are still not well understood, further investigations are warranted.

Our meta-analysis has several limitations. First, because of the cross-sectional design of the included studies, we need to be cautious when making causal inferences about the relationship between sleep duration and youth suicidality. Second, most of the included studies used self-reported questionnaires with a single item measuring sleep duration and 13

ACCEPTED MANUSCRIPT suicidality; thus, recall bias could not be excluded. Third, our study participants were mostly from schools, and adolescents who dropped out of school or were not present at school were not included, which might limit the generalizability of the current dose–response

RI PT

meta-analysis. Fourth, because of only a few studies included, a publication bias test was not utilized in the current study. The potential for publication bias could not be excluded.

SC

Nevertheless, we employed a robust search strategy and included a large-scale sample, which greatly increased the statistical power for detecting the dose–response associations between

M AN U

sleep duration and youth suicidal ideation, plans, and attempts. Moreover, the sensitivity analyses provided solid evidence of the robustness of our findings. In addition, the inclusion of studies from various geographic areas substantially increased the generalizability of our

TE D

findings.

In summary, our meta-analysis provides evidence to support that sleep duration plays an

EP

important role in the development of youth suicidal behaviours. Our findings suggests that

AC C

both long and short sleep duration are significantly associated with an increased incident risk of suicidal ideation and attempts, with the lowest risks of suicidal ideation and attempt at sleep durations of 8–9 h per day. In addition, we noted an 11% reduction in the risk of suicide plans for every 1-h increase in sleep duration. These findings have substantial public health implications and suggest increasing the sleep duration for the prevention of youth suicidal ideation, plans, and attempts. Prospective longitudinal studies with longer follow-up times are required to establish causality and to elucidate the underlying mechanisms. Prospective 14

ACCEPTED MANUSCRIPT studies investigating the relationships between sleep deprivation and suicidal ideation in

AC C

EP

TE D

M AN U

SC

RI PT

healthy and clinical subjects are required.

15

ACCEPTED MANUSCRIPT Practice points 1.

U-shaped associations of sleep duration with the risk of suicidal ideation and attempts are observed, with lowest suicidal ideation and attempts risks at sleep durations of 8–9 h per day. A linear relationship between sleep duration and suicidal plan is found, with an 11%

RI PT

2.

reduction in suicide plans risk for every 1-h increase in sleep duration. 3.

Appropriate sleep duration should be considered as one of influencing factors in

SC

developing preventive strategies of youth suicidal ideation, plans, and attempts.

M AN U

Research agenda Future studies should:

1. Adopt a longitudinal study design to investigate the causality of sleep duration and suicidality and its underlying mechanisms.

2. Conduct prospective studies to examine the association between sleep deprivation and

TE D

suicidal ideation in healthy and clinical subjects. 3. Employ objective instruments such as actigraphs to assess sleep duration to allow more precise estimations of sleep duration.

AC C

EP

4. Develop interventions for targeting the issues of sleep duration in adolescents.

16

ACCEPTED MANUSCRIPT REFERENCES [1] Wasserman D, Cheng QI, Jiang G-X. Global suicide rates among young people aged 15-19. World Psychiatry 2005;4:114-20.

RI PT

[2] Patton GC, Coffey C, Sawyer SM, Viner RM, Haller DM, Bose K, et al. Global patterns of mortality in young people: a systematic analysis of population health data. Lancet

SC

2009;374:881-92.

[3] E.J.L. G, Faravelu C, Nutt DJ, Zohar J, editors. Mood Disorder: Clinical Management

M AN U

and Research Issues. New York, NY: John Wiley & Sons, Ltd; 2005.

[4] Bursztein C, Apter A. Adolescent suicide. Curr Opin Psychiatry 2009;22]:1-6. [5] Kokkevi A, Rotsika V, Arapaki A, Richardson C. Adolescents' self-reported suicide

TE D

attempts, self-harm thoughts and their correlates across 17 European countries. J Child Psychol Psychiatry 2012;53:381-9.

EP

[6] Eaton DK, Kann L, Kinchen S, Shanklin S, Ross J, Hawkins J, et al. Youth risk behavior

AC C

surveillance--United States, 2007. MMWR Surveill Summ 2008;57:1-131. [7] Center for Disease Control and Prevention. Youth Risk Behavior Survey. 2007 Available from: http://www.cdc.yrbss. [8] Nock MK, Green JG, Hwang I, McLaughlin KA, Sampson NA, Zaslavsky AM, et al. Prevalence, correlates, and treatment of lifetime suicidal behavior among adolescents: results from the National Comorbidity Survey Replication Adolescent Supplement. JAMA psychiatry 2013;70:300-10. 17

ACCEPTED MANUSCRIPT [9] Grunbaum JA, Kann L, Kinchen S, Ross J, Hawkins J, Lowry R, et al. Youth risk behavior surveillance--United States, 2003. MMWR Surveill Summ 2004;53:1-96. [10] Randall JR, Nickel NC, Colman I. Contagion from peer suicidal behavior in a

RI PT

representative sample of American adolescents. J Affect Disord 2015;186:219-25. [11] Shaffer D, Gould MS, Fisher P, Trautman P, Moreau D, Kleinman M, et al. Psychiatric

SC

diagnosis in child and adolescent suicide. Arch Gen Psychiatry 1996;53(4):339-48.

[12]Dube SR, Anda RF, Felitti VJ, Chapman DP, Williamson DF, Giles WH. Childhood abuse,

M AN U

household dysfunction, and the risk of attempted suicide throughout the life span: findings from the Adverse Childhood Experiences Study. JAMA 2001;286:3089-96. [13] Brunstein Klomek A, Sourander A, Gould M. The association of suicide and bullying in

TE D

childhood to young adulthood: a review of cross-sectional and longitudinal research findings. Can J Psychiatry 2010;55:282-8.

EP

[14] Aseltine RH, Jr., Schilling EA, James A, Glanovsky JL, Jacobs D. Age variability in the

AC C

association between heavy episodic drinking and adolescent suicide attempts: findings from a large-scale, school-based screening program. J Am Acad Child Adolesc Psychiatry 2009;48:262-70.

[15] Swahn MH, Bossarte RM. Gender, early alcohol use, and suicide ideation and attempts: findings from the 2005 youth risk behavior survey. J Adolesc Health 2007;41]:175-81. [16] Tang TC, Ko CH, Yen JY, Lin HC, Liu SC, Huang CF, et al. Suicide and its association with individual, family, peer, and school factors in an adolescent population in southern 18

ACCEPTED MANUSCRIPT Taiwan. Suicide Life Threat Behav 2009;39:91-102. [17] Hirshkowitz H, Whiton K, Albert SM, Alessi C, Bruni O, DonCarlos L, et al. National Sleep Foundation's sleep time duration recommendations: methodology and results

RI PT

summary. Sleep Health 2015;1:40-3.18. Bartel KA, Gradisar M, Williamson P. Protective and risk factors for adolescent sleep: A meta-analytic review. Sleep Medicine Reviews.

SC

2015;21:72-85.

2011;58:637-47.

M AN U

[19] Carskadon MA. Sleep in adolescents: the perfect storm. Pediatr Clin North Am

[20] Carskadon MA, Vieira C, Acebo C. Association between puberty and delayed phase preference. Sleep 1993;16:258-62.

TE D

[21] Felden ÉPG, Filipin D, Barbosa DG, Andrade RD, Meyer C, Louzada FM. Factors associated with short sleep duration in adolescents. Revista Paulista de Pediatria

EP

2016;34:64-70.

AC C

[22] Chen J, Wan Y, Sun Y, Tao F. [Relations between problems on sleeping and suicidal behaviors in middle school students]. Zhonghua liu xing bing xue za zhi = Zhonghua liuxingbingxue zazhi 2014;35:129-33. [23] Do YK, Shin E, Bautista MA, Foo K. The associations between self-reported sleep duration and adolescent health outcomes: what is the role of time spent on Internet use? Sleep Med 2013;14:195-200. [24] Gangwisch JE, Babiss LA, Malaspina D, Turner JB, Zammit GK, Posner K. Earlier 19

ACCEPTED MANUSCRIPT parental set bedtimes as a protective factor against depression and suicidal ideation. Sleep 2010;33:97-106. [25] Guo L, Xu Y, Deng J, Huang J, Huang G, Gao X, et al. Association between sleep

RI PT

duration, suicidal ideation, and suicidal attempts among Chinese adolescents: the moderating role of depressive symptoms. J Affect Disord 2017;208:355-62.

SC

[26] Jang SI, Lee KS, Park EC. Relationship between current sleep duration and past suicidal ideation or attempt among Korean adolescents. J Prev Med Public Health

M AN U

2013;46:329-35.

[27] Kim JH, Park EC, Lee SG, Yoo KB. Associations between time in bed and suicidal thoughts, plans and attempts in Korean adolescents. BMJ open. 2015;5:e008766.

2012;35:455-60.

TE D

[28] Lee YJ, Cho SJ, Cho IH, Kim SJ. Insufficient sleep and suicidality in adolescents. Sleep

EP

[29] Liu X. Sleep and adolescent suicidal behavior. Sleep 2004;27:1351-8.

AC C

[30] Park JH, Yoo JH, Kim SH. Associations between non-restorative sleep, short sleep duration and suicidality: findings from a representative sample of Korean adolescents. Psychiatry Clin Neurosci 2013;67:28-34. [31] Park TJ, Kim J. Is insufficient sleep duration associated with suicidal behavior in Korean adolescents? Sleep Biol Rhythms. 2017;15:117-25. [32] Winsler A, Deutsch A, Vorona RD, Payne PA, Szklo-Coxe M. Sleepless in Fairfax: the difference one more hour of sleep can make for teen hopelessness, suicidal ideation, and 20

ACCEPTED MANUSCRIPT substance use. J Youth Adolesc 2015;44:362-78. [33] Jia CXL, S.B., Han M, Bo QG. Health-related factors and suicidal ideation in high school students in rural china. J Death Dying 2015:1-12.

RI PT

[34] Fitzgerald CT, Messias E, Buysse DJ. Teen sleep and suicidality: Results from the youth risk behavior surveys of 2007 and 2009. J Clin Sleep Med 2011;7:351-6.

SC

*[35]Pigeon WR, Pinquart M, Conner K. Meta-analysis of sleep disturbance and suicidal thoughts and behaviors. J Clin Psychiatry 2012;73:e1160-7.

M AN U

*[36]Malik S, Kanwar A, Sim LA, Prokop LJ, Wang Z, Benkhadra K, et al. The association between sleep disturbances and suicidal behaviors in patients with psychiatric diagnoses: a systematic review and meta-analysis. Syst Rev 2014;3:18.

TE D

[37]Moher D, Liberati A, Tetzlaff J, Altman DG, Group P. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Int J Surg 2010;8:336-41.

EP

[38] The Joanna Briggs Institute Institute. Joanna Briggs Institute Reviewers' Manual: 2016

AC C

edition. Australia: The Joanna Briggs Institute Institute; 2016 [39] Il’yasova D, Hertz-Picciotto I, Peters U, Berlin JA, Poole C. Choice of exposure scores for categorical regression in meta-analysis: a case study of a common problem. Cancer Causes Control 2005;16:383-8. [40] DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trial 1986;7:177-88. [41] Harrell FE. rms: Regression Modeling Strategies.: R package version 4.5-0; 2016 21

ACCEPTED MANUSCRIPT [Available from: https://CRAN.R-project.org/package=rms. [42] Orsini N, Li R, Wolk A, Khudyakov P, Spiegelman D. Meta-analysis for linear and nonlinear dose-response relations: examples, an evaluation of approximations, and

RI PT

software. Am J Epidemiol 2012;175:66-73. [43] Greenland S, Longnecker MP. Methods for trend estimation from summarized

SC

dose-response data, with applications to meta-analysis. Am J Epidemiol 1992;135:1301-9.

M AN U

[44] Desquilbet L, Mariotti F. Dose-response analyses using restricted cubic spline functions in public health research. Stat Med 2010;29:1037-57.

[45] White IR. Multivariate random-effects meta-analysis. STAT J 2009;9:40-56.

TE D

[46] Orsini N, Bellocco R, Greenland S. Generalized least squares for trend estimation of summarized dose-response data. STAT J 2006;6:40-57.

EP

[47] Hamling J, Lee P, Weitkunat R, Ambuhl M. Facilitating meta-analyses by deriving

AC C

relative effect and precision estimates for alternative comparisons from a set of estimates presented by exposure level or disease category. Stat Med 2008;27:954-70. [48] Jang SI, Lee KS, Park EC. Relationship between current sleep duration and past suicidal ideation or attempt among korean adolescents. J Prev Med Public Health 2013;46:329-35. *[49]Perlis ML, Grandner MA, Chakravorty S, Bernert RA, Brown GK, Thase ME. Suicide and sleep: is it a bad thing to be awake when reason sleeps? Sleep Med Rev 2016;29:101-7. 22

ACCEPTED MANUSCRIPT *[50]Kohyama J. Sleep, serotonin, and suicide in Japan. J Physiol Anthropol 2011;30:1-8. [51] Leyton M, Paquette V, Gravel P, Rosa-Neto P, Weston F, Diksic M, et al. alpha-[11C]Methyl-L-tryptophan trapping in the orbital and ventral medial prefrontal

RI PT

cortex of suicide attempters. Eur Neuropsychopharmacol 2006;16:220-3. *[52]Mann JJ. The serotonergic system in mood disorders and suicidal behaviour. Philos

SC

Trans R Soc Lond B Biol Sci 2013;368:20120537.

[53] Underwood MD, Kassir SA, Bakalian MJ, Galfalvy H, Mann JJ, Arango V. Neuron

M AN U

Density and Serotonin Receptor Binding in Prefrontal Cortex in Suicide. Int J Neuropsychopharmacol 2012;15:435-47.

[54] Portas CM, Bjorvatn B, Ursin R. Serotonin and the sleep/wake cycle: special emphasis on

TE D

microdialysis studies. Progress in neurobiology. 2000;60(1):13-35. [55] Roman V, Walstra I, Luiten PG, Meerlo P. Too little sleep gradually desensitizes the

EP

serotonin 1A receptor system. Sleep 2005;28:1505-10.

AC C

[56] Novati A, Roman V, Cetin T, Hagewoud R, den Boer JA, Luiten PG, et al. Chronically restricted sleep leads to depression-like changes in neurotransmitter receptor sensitivity and neuroendocrine stress reactivity in rats. Sleep 2008;31:1579-85. [57] Kim JH, Park EC, Cho WH, Park CY, Choi WJ, Chang HS. Association between total sleep duration and suicidal ideation among the Korean general adult population. Sleep 2013;36:1563-72.

23

Table 1 Metaregression and subgroup analyses ACCEPTED Factors

MANUSCRIPT

I2 (%)

Point (95% CI)

n

P for heterogeneity*

estimate Suicidal ideation (n = 13) Metaregression Mean age

8

-0.29 (-0.59 to 0.003)

85.6

Female percentage

13

0.0002 (-0.005 to 0.006)

92.8

RI PT

Subgroup analyses Sample size

0.19

≥20,000

8

0.52 (0.42 to 0.63)

94.9

<20,000

5

0.64 (0.53 to 0.78)

55.4

Study region

0.33

3

0.58 (0.40 to 0.84)

85.2

South Korea

7

0.53 (0.43 to 0.66)

95.5

USA

3

0.59 (0.40 to 0.87)

76.3

Yes

6

No

7

Statistical adjusted for depression 8

No

5

Statistical adjusted for smoking Yes No

Yes No

0.76

0.57 (0.49 to 0.66)

85.4

0.54 (0.43 to 0.69)

91.0 0.14

0.60 (0.51 to 0.70)

88.1

0.48 (0.37 to 0.62)

89.3 0.71

7

0.54 (0.45 to 0.64)

90.5

6

0.58 (0.43 to 0.78)

92.1

EP

Statistical adjusted for drinking

TE D

Yes

M AN U

Weekday sleep duration

SC

China

0.71

7

0.54 (0.45 to 0.64)

90.5

6

0.58 (0.43 to 0.78)

92.1

Yes No

AC C

Statistical adjusted for drug use

0.36

3

0.49 (0.38 to 0.63)

85.7

10

0.58 (0.48 to 0.70)

93.5

Statistical adjusted for other sleep variables†

0.13

Yes

6

0.62 (0.51 to 0.76)

89.7

No

7

0.50 (0.41 to 0.61)

88.5

Mean age

5

0.016 (-0.52 to 0.54)

0

Female percentage

6

0.001 (-0.008 to 0.011)

0

Suicidal plan(n = 6) Metaregression

Subgroup analyses Sample size

0.19

3 0.41 (0.32 to 0.53) ACCEPTED MANUSCRIPT

≥20,000 <20,000

3

0.52 (0.46 to 0.59)

0 0

Study region

0.64

China and South Korea

4

0.49 (0.43 to 0.56)

2.4

America

2

0.54 (0.35 to 0.81)

0

Weekday sleep duration

0.46 2

0.50 (0.44 to 0.60)

0

No

4

0.51 (0.45 to 0.57)

0

RI PT

Yes

0.19

Statistical adjusted for depression Yes

3

0.41 (0.32 to 0.53)

0

No

3

0.52 (0.46 to 0.59)

0

Statistical adjusted for smoking

SC

0.19

3

0.41 (0.32 to 0.53)

0

No

3

0.52 (0.46 to 0.59)

0

Statistical adjusted for drinking Yes

3

No

3

Statistical adjusted for other sleep †

variables

3

No Suicidal attempt (n = 11) Metaregression

Female percentage Subgroup analyses

AC C

Sample size

0.41 (0.32 to 0.53)

0

0.52 (0.46 to 0.59)

0 0.19 0

3

0.52 (0.46 to 0.59)

0

7

0.38 (-0.097 to 0.852)

47.2

11

0.003 (-0.007 to 0.012)

80.6

EP

Mean age

0.19

0.41 (0.32 to 0.53)

TE D

Yes

M AN U

Yes

0.82

≥20,000

7

0.53 (0.38 to 0.73)

85.4

<20,000

4

0.50 (0.41 to 0.61)

0

Study regions China

0.80 3

0.58 (0.39 to 0.87)

77.2

South Korea

6

0.49 (0.32 to 0.74)

86.3

America

2

0.56 (0.32 to 0.96)

5.4

Weekday sleep duration

0.65

Yes

6

0.49 (0.34 to 0.70)

87.3

No

5

0.55 (0.42to 0.73)

21.6 0.44

Adjusted for depression Yes

7

0.47 (0.32 to 0.70)

83.9

No

4

0.60 (0.44 to 0.83)

66.2

ACCEPTED MANUSCRIPT

Statistical adjusted for smoking

0.82

Yes

7

0.50 (0.41 to 0.61)

85.4

No

4

0.53 (0.38 to 0.73)

78.5

Statistical adjusted for drinking

0.65

Yes

6

0.49 (0.34 to 0.70)

87.3

No

5

0.55 (0.42 to 0.73)

21.6

Statistical adjusted for drug use

0.95 3

0.51 (0.31 to 0.84)

71.1

No

8

0.52 (0.39 to 0.70)

81.0

RI PT

Yes

Statistical adjusted for other sleep

0.97



variables

5

0.51 (0.30 to 0.86)

No

6

0.53 (0.40 to 0.69)

SC

Yes

68.5 86.4

M AN U

CI = confidence interval. * Between subgroups with metaregression analysis. † perception of getting enough sleep, awakening time and bedtime, nightmares and insomnia, restorative sleep,

AC C

EP

TE D

and chronotype.

ACCEPTED MANUSCRIPT

RI PT

Potentially relevant articles identified (n = 640) 33 EMBASE 34 PubMed 1 PsycINFO 567 ProQuest Dissertations & Theses A&I 4 Wanfang Data Chinese database 0 China Knowledge Resource Integrated Database 1 Hand searching by checking the reference lists of included studies

SC

Articles excluded after screening of title and abstract (n = 627) - Did not report the sleep duration - Did not mention suicidality

M AN U

Full text articles extracted for detailed evaluation (n = 13) Excluded (n = 3) -Did not report sleep duration category

TE D

10 eligible articles were included and 1 article had rate ratios in different years (i.e., 2007 and 2009) and 2 exclusively involved gender specific rate ratios, resulting in 13 eligible reports for systematic review

EP

Articles included in systematic review (n = 13) Suicidal ideations (n = 13) Suicidal plans (n = 6) Suicidal attempts (n = 11)

AC C

Articles with only 2 sleep duration categories (n = 1) was excluded from dose-response metaanalysis

Articles included in dose-response metaanalysis (n = 12) Suicidal ideations (n = 12) Suicidal plans (n = 5) Suicidal attempts (n = 10)

Figure 1 Flow chart of literatures search for studies investigating association between sleep duration and the risks of suicidal ideations, plans, and attempts.

ACCEPTED MANUSCRIPT

(B)

TE D EP

Figure 2 the linear and non-linear associations of sleep duration with (A) suicidal ideation, (B) suicidal plan, (C) suicidal attempt.

AC C

(C)

M AN U

SC

RI PT

(A)