Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 Aug 1.
Published in final edited form as: Psychiatry Res. 2023 Jun 25;326:115323. doi: 10.1016/j.psychres.2023.115323

Examining sleep disturbance components as near-term predictors of suicide ideation in daily life

Rebecca C Cox a,b, Sarah L Brown b, Brittany N Chalmers b, Lori N Scott b,⁎
PMCID: PMC10527974  NIHMSID: NIHMS1912688  PMID: 37392522

Abstract

Suicide ideation emerges and fluctuates over short timeframes (minutes, hours, days); however, near-term predictors of such fluctuations have not been well-elucidated. Sleep disturbance is a distal suicide risk factor, but less work has examined whether daily sleep disturbance predicts near-term changes in suicide ideation. We examined subjective sleep disturbance components as predictors of passive and active suicide ideation at the within-person (i.e., day-to-day changes within individuals relative to their own mean) and between-persons (individual differences relative to the sample mean) levels. A transdiagnostic sample of 102 at-risk young adults ages 18–35 completed a 21-day ecological momentary assessment protocol, during which they reported on sleep and passive and active suicide ideation. At the within-persons level, nightmares, sleep quality, and wake after sleep onset predicted passive suicide ideation, and sleep quality and wake after sleep onset predicted active suicide ideation. At the between-persons level, nightmares, sleep onset latency, and sleep quality were associated with passive suicide ideation, and sleep onset latency was associated with active suicide ideation. In contrast, suicide ideation did not predict subsequent sleep at the within-person level. Specific sleep disturbance components are near-term predictors of intraindividual increases in suicide ideation and may hold promise for suicide prevention and intervention.

Keywords: Ecological momentary assessment, Suicide risk, Sleep, Nightmares, Daily

1. Introduction

Suicide is the second leading cause of death among young adults (Centers for Disease Control [CDC], 2023). In 2018, 4.3% of adults reported “serious thoughts of suicide in the past year,” with the highest rates (11%) among young adults 18–25 (Substance Abuse and Mental Health Services Administration, 2019). Despite decades of research, our ability to predict, prevent, and treat suicide risk remains inadequate (Brown and Jager-Hyman, 2014; Fox et al., 2020; Franklin et al., 2017; Riblet et al., 2017). Though numerous correlates and distal (i.e., further away in time and/or more indirect associations) risk factors for suicide have been identified (e.g., Franklin et al. 2017), less is known about proximal (i.e., closer in time and/or more direct associations) predictors for acute fluctuations in suicide risk among vulnerable individuals. Recent evidence suggests suicide ideation (SI) and suicidal behavior, as well as their precipitants, emerge and fluctuate in a matter of minutes, hours, and days (Coppersmith et al., 2019; Hallensleben et al., 2018; Kleiman et al., 2017; Millner et al., 2017; Nock et al., 2009). Intensive longitudinal studies with at-risk individuals are necessary to answer critical questions that can guide prevention and intervention efforts: which factors predict acute increases in suicide risk, and when an individual is at elevated risk.

Sleep disturbance is a promising modifiable risk factor for acute changes in suicide risk. There is robust evidence that sleep disturbance is cross-sectionally and prospectively associated with elevated risk for SI, attempts, and deaths by suicide (Liu et al., 2019; Pigeon et al., 2012). Previous research has found multiple aspects of sleep disturbance are linked to elevated SI, including insomnia symptoms (Chan et al., 2020; Dolsen et al., 2021), both short and long sleep duration (Hedström et al., 2021; Khader et al., 2020), nocturnal wakefulness (Tubbs et al., 2021), and nightmares (Becker et al., 2018; Russell et al., 2018). Importantly, sleep disturbance is associated with elevated risk for SI and suicidal behaviors over and above the effects of depression (Batterham et al., 2021; Nardorff et al., 2013; Pigeon et al., 2012), anxiety, posttraumatic stress disorder (Bishop et al., 2020; Nardorff et al., 2013), substance use disorder, bipolar disorder, schizophrenia, and medical comorbidities (Bishop et al., 2020), suggesting that sleep disturbance uniquely contributes to suicide risk. Further, sleep disturbance prospectively predicts SI and suicidal behaviors over days to years (Liu et al., 2020). Indeed, nightmares (Liu et al., 2021), short sleep duration, and poor sleep quality (Gong et al., 2020) predict increased risk for SI, suicide plans, and suicide attempts over 1- and 2- years. Another study found that insomnia symptoms predicted elevated risk for SI and suicide attempts one month later, adjusting for depression and hopelessness (Ribeiro et al., 2012). Taken together, previous research has established sleep disturbance as a robust correlate and distal predictor of SI and suicidal behaviors. This literature supports the development of the “mind after midnight” hypothesis, which proposes that insufficient or disrupted sleep results in excessive wakefulness during the night, when the brain is vulnerable to putative mechanisms of SI and suicidal behaviors, such as negative affect, executive dysfunction, and attentional biases (Tubbs et al., 2022).

Although sleep and suicide risk are dynamic processes that unfold over relatively short time frames, and sleep is conceptualized as an acute risk factor for suicide, less is known about whether nightly sleep disturbance contributes to near-term increases in suicide risk. There has been a recent call for research examining sleep disturbance as a near-term risk factor for suicide (Liu et al., 2020). Studies using ecological momentary assessment (EMA) to sample daily variability in sleep and SI and suicidal behaviors may yield novel insight into the relations between components of sleep disturbance and acute fluctuations in suicide risk (Sedano-Capdevila et al., 2021). This research is necessary to determine which components of sleep disturbance would be most appropriate as targets for prevention or intervention. Additionally, EMA methods offer decreased retrospective bias and enhanced real-world validity (Shiffman et al., 2008), which may clarify the complex associations between components of sleep disturbance and suicide risk (Gratch et al., 2021). Finally, EMA methods offer the opportunity to disentangle which individuals are at risk for passive or active SI and near-term elevations in passive or active SI within the individual.

Despite this potential, few studies to date have used EMA to examine the relations between daily sleep disturbance and suicide risk, and findings are inconsistent. In a study of adults with current SI, decreased objective and subjective sleep duration and decreased sleep quality, but not sleep efficiency or sleep onset latency, predicted increased next-day SI (Littlewood et al., 2019). In contrast, in a sample of adolescents recently discharged from acute care for suicide risk, longer subjective sleep onset latency and more nightmares were associated with increased next-day SI, whereas no effect of objective nor subjective sleep duration was found (Glenn et al., 2021). Surprisingly, better sleep quality and lower objective wake after sleep onset were also associated with increased next-day SI (Glenn et al., 2021). Similarly, a recent study of adolescents and young adults undergoing intensive outpatient treatment for depression and suicide risk found no direct effect of sleep duration or sleep quality on next-day SI, though indirect effects through affective reactivity to interpersonal events were observed (Hamilton et al., 2022). Another recent study of psychiatric inpatient adults with depression found that general sleep disturbance predicted active, but not passive, next-day SI (Brüdern et al., 2022); however, this study did not distinguish between specific components of sleep disturbance. Finally, a study of adults with borderline personality disorder found no effect of last night’s sleep on next-day SI, although those who took longer to fall asleep reported higher SI on average (Kaurin et al., 2021).

Given that the EMA literature on daily sleep and SI is limited and inconsistent, additional EMA studies are needed to clarify the near-term associations between components of sleep disturbance and SI among young adults. Further, few previous EMA studies have separately examined passive (i.e., thoughts of death) and active (i.e., thoughts of ending one’s own life) SI. This is important considering these are theoretically distinct (Klonsky and May, 2015; Van Orden et al., 2010), have differential associations with predictors and clinical outcomes (Jahn et al., 2018; O’Riley et al., 2014), and reflect different levels of severity that may impact clinical care and management (Chu et al., 2015). Finally, although subjective sleep measures can be vulnerable to retrospective bias, they are more feasible for immediate clinical implementation; thus, clarifying subjective sleep predictors of daily SI may provide more clinical utility for identifying acute SI risk in real-world settings.

The present study sought to address these gaps in the literature using EMA with a transdiagnostic sample of young adults with a recent history of SI or suicidal behaviors. We first examined subjective sleep disturbance components (i.e., increased sleep onset latency, increased wake after sleep onset, decreased sleep duration, later sleep timing, decreased sleep quality, and increased nightmares) as separate predictors of passive and active SI at the within- and between-person levels. Next, we examined which sleep disturbance components uniquely predicted passive and active SI within- and between-persons, over and above other sleep disturbance components. We hypothesized that specific components of daily sleep disturbance would be independently and uniquely: (1) associated with elevated risk for passive and active SI between-persons, and (2) predict within-person increases in passive and active SI the following day. Given prior inconsistent findings, we did not hypothesize which components of sleep disturbance would emerge as significant predictors. Exploratory reverse models with passive and active SI predicting near-term within-person changes in sleep disturbance components were tested to examine bidirectional effects.

2. Methods

2.1. Participants

Participants were 102 young adult men and women between the ages of 18 and 35 (Mage=24.92, SDage=4.68) with SI and/or suicidal behaviors (i.e., preparatory behaviors and/or aborted, interrupted, or actual suicide attempts per the Columbia Suicide Severity Rating Scale, Posner et al. 2011) within the past 4 months. See Table 1 for demographics and diagnostic information.

Table 1.

Demographic and clinical diagnostic information.

n (%)
Gender
Male 16 (15.7)
Female 76 (74.5)
Transgender Female-to-male 2 (2.0)
Transgender Male-to-female 1 (1.0)
Non-binary or other 7 (6.9)
Ethnicity
Non-Hispanic/LatinX 90 (88.2)
Hispanic/LatinX 10 (9.8)
Race
Asian 9 (8.8)
Black or African American 18 (17.6)
White 66 (64.7)
Multiracial 7 (6.9)
Sexual Orientation
Heterosexual/straight 50 (49.0)
Gay/lesbian/homosexual 9 (8.8)
Bisexual 27 (26.5)
Other/Unsure 14 (13.7)
Education
Less than Highschool or GED 1 (1.0)
Highschool or GED 8 (7.8)
Some College 36 (35.3)
Technical/trade School Certificate 3 (2.9)
College Degree or Higher 54 (52.9)
Employment
Not employed 27 (26.4)
Part-time 38 (37.3)
Full-time 37 (36.3)
Annual Income: M (SD) $71,594 ($67,744)
C-SSRS Suicide Risk
NSSI history 70 (68.6)
Suicide attempt history 39 (38.2)
Multiple suicide attempt history 17 (16.8)
SI Intensity: M (SD) 14.47 (4.06)
SCID-5-RV Lifetime Diagnosis
Mood Disorder 102 (100)
Anxiety Disorder 82 (80.4)
Obsessive-compulsive Disorder 22 (21.6)
Trauma and Stress-related Disorder 50 (59.0)
Substance Use Disorder 58 (56.9)
Eating Disorder 32 (31.4)
SIDP-IV Current Diagnoses
Borderline Personality Disorder 14 (13.7)
Antisocial Personality Disorder 3 (2.9)
Avoidant Personality Disorder 24 (23.5)
Open in a new tab

Note. NSSI = non-suicidal self-injury. Suicide Risk was assessed using the Columbia Suicide Severity Rating Scale (C-SSRS;Posner et al., 2011). SI Intensity = suicide ideation intensity for worst level suicide ideation in the past 4 months. DSM-V Lifetime diagnoses were assessed using the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (First et al., 2015). Selected personality disorders were assessed using relevant items from the Structured Clinical Interview for DSM-IV Personality Disorders (Pfohl et al., 1997), the criteria for which are unchanged in DSM-V. Percentages may not sum to 100% due to missing data.

2.2. Recruitment

Participants were recruited for an ongoing one-year longitudinal study of suicide risk in a transdiagnostic sample of young adults. Participants were eligible if they had experienced active SI at least two times in a week or suicidal behaviors in the past 4 months and were receiving mental health care at least two times a month at initial recruitment. Participants completed diagnostic interviews, laboratory assessments, and self-report questionnaires at baseline, 4-, 8-, and 12-month followup assessments. Additionally, participants completed a 21-day web-based EMA protocol on smartphones following baseline. Only the baseline diagnostic interview and EMA data were examined. All participants provided written and verbal informed consent, and study procedures were conducted in accordance with the responsible Institutional Review Board.

2.3. EMA protocol

Seven semi-random surveys per day were scheduled between participants’ self-reported wake and sleep schedules for a maximum of 147 possible prompts over 21 days. Surveys included wake-up assessment items (once daily), momentary items (at all prompts), items capturing experiences since the last assessment (at all prompts), and end-of-day items capturing experiences across the entire day (once daily). On average, participants completed 126 prompts (SD = 27.27; Range = 14–149; protocol was extended up to 7 days for 3 participants who experienced technical difficulty; mean compliance = 86%). Data were monitored regularly for risk assessment and compliance. Participants were paid up to $175 plus a $20 bonus for completing at least 85% of prompts over the 21-day EMA period. Compensation was provided weekly (week 1 = $50, week 2 = $60, week 3 = $70), and completion rates less than 85% earned a prorated amount.

2.4. Measures

2.4.1. Daily passive and active suicide ideation

For each prompt, participants reported whether they had experienced SI since the last completed assessment on two yes/no items adapted for self-report from the Columbia Suicide Severity Rating Scale (Posner et al., 2011) assessing passive ideation (“I wished to be dead or wished I could go to sleep and not wake up”) and active SI (“I thought about killing myself”). These binary variables were then collapsed to the day level, and two binary variables were created and coded as “1″ to indicate if the participant endorsed either passive or active SI on a given day, respectively, on at least one of the EMA prompts throughout the day, and 0 indicating the participant responded “no” to ideation on all the EMA prompts on a given day. We maintained the binary nature of the variables to limit the impact of skewness. Most participants endorsed passive (n=71, 70%) and active SI (n=61, 60%) at least one time. On average, participants endorsed passive SI on 4.99 (SD=3.74) days and active SI on 2.53 (SD=3.74) days.

2.4.2. Daily sleep disturbance

Participants reported on the previous night’s sleep during their first assessment of the day using items from the Pittsburgh Sleep Diary (Monk et al., 1994). Sleep diaries are the gold-standard method for subjective sleep assessment (Carney et al., 2012). See Table 2 for sleep variables assessed and their definitions/calculations.

Table 2.

Description of sleep variables.

Sleep variable Definition Calculation
Bedtime Clock time lights were turned off for bed
Sleep onset latency Minutes to fall asleep
Sleep onset Clock time sleep was initiated Lights out time plus sleep onset latency
Number of awakenings Number of awakenings between sleep onset and offset
Wake after sleep onset Minutes awake between sleep onset and sleep offset
Sleep offset Clock time of final awakening
Sleep duration Duration of the total sleep period in minutes (Sleep onset minus sleep offset) minus wake after sleep onset
Sleep timing Clock time of the midpoint between sleep onset and sleep offset
Sleep quality Rating of sleep quality on visual analogue scale; 0 Very Bad, 100 Very Good
Nightmares Number of unpleasant dreams/nightmares
Open in a new tab

Prior to calculating sleep duration and sleep timing, we inspected bedtime and waketime for errors. We corrected 11 instances where AM/PM were clearly misreported (e.g., participant reported bedtime 11:55am and waketime 8:00am, bedtime was changed to 11:55pm). Ten instances of apparent errors in bedtime and/or wake time without a plausible correction were removed (e.g., participant reported bedtime 4:00am and waketime 4:00am). After calculating sleep duration and sleep timing, we removed 13 records with negative sleep duration values (e.g., participant reported bedtime 2:20am, waketime 4:00am, sleep onset latency 120 min, and wake after sleep onset 300 min).

2.4.3. Daily depressed mood

At each EMA prompt, participants rated the following statement on a scale of 0 (Not at all) to 100 (Very much): “At this moment I feel…depressed.” These seven daily momentary assessments were averaged to create a daily depressed mood score.

2.5. Data analytic strategy

Hypotheses were tested using 2-level multilevel models with fixed effects using Bayes estimation in Mplus version 8.5 (Muthen and Muthen, 2020), which has been shown to perform well with small samples and categorical outcomes (McNeish, 2016). Momentary data (7 prompts daily) were collapsed to the day level, such that 21 days (level 1) were nested within 102 participants (level 2). Importantly, to examine within-individual change in the likelihood of SI relative to the previous day’s SI, we included previous day’s passive and active SI as predictors of current-day ideation. Variables with both within- and between-person variance were person-mean-centered at Level 1 (daily within-person changes in sleep variables, lagged daily depressed mood, and lagged passive and active SI) and grand-mean-centered at level 2 (each person’s average sleep variables, depressed mood, and SI) using automatic latent mean centering in Mplus. Thus, within-person effects can be interpreted as fluctuations relative to a person’s own average levels, and between-person effects can be interpreted as individual differences in overall levels across the assessment period relative to the sample mean. As an exploratory analysis of bidirectional effects, we also examined reverse models with daily passive and active SI predicting the subsequent night’s sleep disturbance components, controlling for previous night’s sleep disturbance. Reported estimates are standardized posterior mean point estimates and their corresponding posterior standard deviations and 95% credibility intervals (CIs).

We tested associations between demographic variables and passive and active SI at the participant level to identify potential covariates. Point biserial correlations between passive and active SI and age and chi-square tests of associations between passive and active SI and gender, sexual orientation, marital status, education level, household income, and religion revealed no significant associations (p’s>0.05). Therefore, no demographic variables were included as covariate predictors in the models. Likewise, a chi-square test revealed that participation before or during the COVID-19 pandemic (coded 0 and 1 for before and during COVID-19, respectively) was also not significantly associated with passive or active SI (p’s>0.05) and therefore was not included in the model. Depressed mood was included as a covariate given the well-established relation between depressed mood and SI (e.g., Franklin et al. 2017) and the moderate, significant bivariate associations observed between depressed mood and passive and active SI in the present sample1. To control for trends over time in outcomes, we included time elapsed since the first EMA assessment (“days passed,” person-mean-centered) as a within-person (level 1) predictor in all models.

All models were tested with Bayesian Markov chain Monte Carlo (MCMC) estimation, default (i.e., noninformative) priors, 20,000 iterations, and 2 MCMC chains. We first tested each sleep disturbance component in separate models. Sleep disturbance components that were significant predictors of passive or active SI in the individual models were then included in an omnibus model to examine their unique effects. We examined proportional scale reduction (PSR) values to assess model convergence, where values close to 1 indicate convergence (Asparouhov and Muthén, 2010). Across all models, the highest PSR value observed was 1.01 indicating satisfactory convergence. Models were then re-tested with iterations increased to 40,000, which did not result in increased PSR values, further indicating acceptable convergence. Examination of variance inflation factors (VIF) of sleep disturbance components prior to testing the omnibus model revealed no evidence of multi-collinearity (VIFs>1.20 and <1.72); thus, these variables could be examined simultaneously.

3. Results

3.1. Descriptive statistics and bivariate correlations

Descriptive statistics and associations between average sleep disturbance, depressed mood, and SI are shown in Table 3.

TABLE 3.

Descriptive statistics and bivariate correlations at the between-person level (N = 102).

Variable 1 2 3 4 5 6 7 8 9 10
1. Passive SI –
2. Active SI 0.46** –
3. SOL 0.27** 0.19 –
4. WASO 0.32** 0.15 0.35** –
5. Sleep duration −0.10 0.09 −0.18 −0.32** –
6. Sleep timing −0.04 0.02 0.32** −0.11 0.07 –
7. Sleep quality −0.35** −0.04 −0.51** −0.49** 0.37** −0.17 –
8. Nightmares 0.27** 0.12 0.24* 0.38** −0.11 0.06 −0.29** –
9. Depressed 0.34** 0.27** 0.24* 0.31** −0.16 0.10 −0.33** 0.25* –
10. Days passed 0.12 0.07 0.04 0.06 −0.09 −0.14 −0.18 −0.02 −0.22* –
M/% 70% 60% 22.5 19.49 444.97 4.33 51.42 0.39 29.24 10.15
SD 24.9 36.65 115.20 1.69 26.62 0.77 25.36 6.16
Range 0–120 0–300 15–1024.8 −1.33–16.46 0–100 0–7 0–100 0–27.25
Open in a new tab

Note. SI = suicide ideation (coded as 0 = no, 1= yes); SOL = sleep onset latency; WASO = wake after sleep onset; Depressed = depressed mood; Days passed = time elapsed since the first ecological momentary assessment (EMA) assessment. Percentages for passive and active SI reflect percent of the sample that reported passive and active suicide ideation at least once during the EMA period. SOL, WASO, and sleep duration are reported in minutes. Sleep timing is reported in decimal format (e.g., 4.37 = 4:22 am). Sleep quality was reported from 0 (Very Bad) to 100 (Very Good). Nightmares were reported as the discrete number of nightmares the previous night. Depressed mood was reported from 0 (Not at all) to 100 (Very much).

*

p < .05

**

p < .01

3.2. Individual models

3.2.1. Sleep onset latency

Results revealed no significant effect of daily fluctuations in sleep onset latency on the likelihood of next-day passive or active SI at the within-person level. However, there were significant, positive effects of sleep onset latency on passive and active SI at the between-persons level, such that participants who took longer to fall asleep on average were more likely to experience passive and active SI during the sampling period (see Table 4).

TABLE 4.

Sleep disturbance components as separate predictors of passive and active suicide ideation.

Passive SI
 Active SI
Fixed effects Predictor Est Posterior SD 95% CI (lower, upper) Est Posterior SD 95% CI (lower, upper)
Main Predictor: SOL
Level 1 Lagged passive SI 0.41 0.11 0.22, 0.66 0.24 0.10 0.06, 0.46
Lagged active SI 0.16 0.09 −0.01, 0.35 0.33 0.11 0.11, 0.54
Lagged depressed mood −0.03 0.04 −0.11, 0.05 0.02 0.05 −0.07, 0.12
Days passed <−0.01 0.03 −0.07, 0.06 −0.12 0.04 −0.20, −0.04
SOL 0.03 0.03 −0.04, 0.09 0.04 0.04 −0.03, 0.11
Level 2 Depressed mood 0.57 0.10 0.37, 0.76 0.30 0.12 0.06, 0.54
SOL 0.22 0.10 0.02, 0.42 0.24 0.12 0.01, 0.49
Residual variance 0.62 0.11 0.35, 0.80 0.83 0.09 0.62, 0.96
Main Predictor: WASO
Level 1 Lagged passive SI 0.44 0.13 0.23, 0.73 0.27 0.12 0.07, 0.54
Lagged active SI 0.18 0.10 0.001, 0.38 0.35 0.11 0.12, 0.57
Lagged depressed mood −0.05 0.04 −0.13, 0.04 <0.01 0.05 −0.08, 0.11
Days passed <−0.01 0.03 −0.07, 0.06 −0.12 0.04 −0.20, −0.04
WASO 0.06 0.03 0.004, 0.12 0.07 0.04 0.003, 0.15
Level 2 Depressed mood 0.57 0.12 0.36, 0.83 0.32 0.14 0.05, 0.60
WASO 0.22 0.12 −0.01, 0.48 0.11 0.15 −0.18, 0.40
Residual variance 0.62 0.14 0.24, 0.81 0.86 0.10 0.58, 0.98
Main Predictor: Sleep Duration
Level 1 Lagged passive SI 0.44 0.13 0.24, 0.73 0.27 0.12 0.07, 0.53
Lagged active SI 0.17 0.10 −0.01, 0.36 0.33 0.11 0.12, 0.56
Lagged depressed mood −0.04 0.05 −0.12, 0.05 0.01 0.05 −0.08, 0.11
Days passed <−0.01 0.03 −0.07, 0.06 −0.12 0.04 −0.20, −0.04
Sleep duration <−0.01 0.03 −0.07, 0.06 −0.03 0.04 −0.11, 0.05
Level 2 Depressed mood 0.60 0.11 0.40, 0.84 0.36 0.13 0.10, 0.62
Sleep duration −0.12 0.11 −0.37, 0.08 −0.02 0.14 −0.31, 0.23
Residual variance 0.62 0.14 0.24, 0.82 0.86 0.10 0.59, 0.98
Main Predictor: Sleep Timing
Level 1 Lagged passive SI 0.41 0.11 0.22, 0.67 0.24 0.11 0.06, 0.48
Lagged active SI 0.14 0.10 −0.04, 0.34 0.30 0.11 0.08, 0.52
Lagged depressed mood −0.04 0.04 −0.11, 0.05 0.03 0.05 −0.07, 0.12
Days passed −0.01 0.03 −0.08, 0.06 −0.13 0.04 −0.21, −0.05
Sleep timing −0.05 0.03 −0.12, 0.01 −0.06 0.04 −0.14, 0.03
Level 2 Depressed mood 0.60 0.09 0.41, 0.80 0.35 0.12 0.11, 0.57
Sleep timing 0.05 0.10 −0.14, 0.25 −0.03 0.12 −0.27, 0.22
Residual variance 0.63 0.12 0.35, 0.82 0.86 0.09 0.65, 0.98
Main Predictor: Sleep Quality
Level 1 Lagged passive SI 0.38 0.11 0.19, 0.62 0.20 0.10 0.02, 0.43
Lagged active SI 0.18 0.10 0.00, 0.38 0.36 0.11 0.14, 0.58
Lagged depressed mood −0.04 0.04 −0.12, 0.05 0.02 0.05 −0.07, 0.12
Days passed −0.01 0.03 −0.08, 0.06 −0.12 0.04 −0.20, −0.05
Sleepquality −0.16 0.04 −0.23, −0.09 −0.08 0.04 −0.16, −0.001
Level 2 Depressed mood 0.53 0.10 0.32, 0.72 0.34 0.13 0.08, 0.57
Sleep quality −0.23 0.11 −0.46, −0.02 −0.03 0.14 −0.31, 0.22
Residual variance 0.65 0.10 0.40, 0.82 0.87 0.09 0.65, 0.98
Main Predictor: Nightmares
Level 1 Lagged passive SI 0.41 0.11 0.22, 0.65 0.22 0.10 0.04, 0.45
Lagged active SI 0.16 0.10 −0.02, 0.36 0.33 0.11 0.11, 0.56
Lagged depressed mood −0.04 0.04 −0.12, 0.05 0.02 0.05 −0.07, 0.12
Days passed <−0.01 0.03 −0.07, 0.06 −0.12 0.04 −0.20, −0.05
Nightmares 0.15 0.03 0.08, 0.21 0.07 0.04 −0.001, 0.14
Level 2 Depressed mood 0.56 0.10 0.37, 0.77 0.34 0.12 0.09, 0.57
Nightmares 0.26 0.11 0.04, 0.47 <−0.01 0.14 −0.27, 0.26
Residual variance 0.61 0.12 0.30, 0.79 0.87 0.09 0.64, 0.98
Open in a new tab

Note. SI = suicide ideation; SOL = sleep onset latency; WASO = wake after sleep onset; Days passed = time elapsed since the first ecological momentary assessment (EMA) . Level 1 = within-person fluctuations at the daily level. Level 2 = between-persons differences across the assessment period. Standardized point estimates (Est) are reported. Bolded estimates are significant (i.e., 95% CIs do not cross zero).

3.2.2. Wake after sleep onset

Results revealed significant, positive effects of wake after sleep onset on passive and active SI at the within-person level, such that nights with more time awake between sleep onset and sleep offset than typical for the individual were followed by days with an increased likelihood of experiencing passive and active SI. In contrast, there were no significant effects of wake after sleep onset on passive or active SI at the between-persons level (see Table 4).

3.2.3. Sleep duration

Results revealed no significant effects of sleep duration on passive or active SI at either the within- or between-persons level (see Table 4).

3.2.4. Sleep timing

Results revealed no significant effects of sleep timing on passive or active SI at either the within- or between-persons level (see Table 4).

3.2.5. Sleep quality

Results revealed significant, negative effects of sleep quality on passive and active SI at the within-person level, such that nights with worse sleep quality than typical for the individual were followed by days with an increased likelihood of passive and active SI. There was also a significant, negative effect of sleep quality on passive SI at the between-persons level, indicating that participants who experienced worse sleep quality on average were more likely to experience passive SI during the sampling period. In contrast, there were no significant effects of sleep quality on active SI at the between-persons level (see Table 4).

3.2.6. Nightmares

Results revealed a significant, positive effect of nightmares on passive SI at the within-person level, indicating that nights with more nightmares than typical for the individual were followed by days with an increased likelihood of passive SI. There was also a significant, positive effect of nightmares on passive SI at the between-persons level, such that participants who experienced more nightmares on average were more likely to experience passive SI during the sampling period. In contrast, there were no significant effects of nightmares on active SI at either the within or between-persons level (see Table 4).

3.3. Omnibus model

We tested an omnibus model that included significant predictors from the individual models (i.e., sleep onset latency, wake after sleep onset, sleep quality, and nightmares) as simultaneous predictors of passive and active SI. Within-person effects of nightmares and sleep quality on passive SI remained significant. However, the within-person effects of sleep quality on active SI and wake after sleep onset on passive and active SI were no longer significant when accounting for other sleep disturbance components (see Table 5).

Table 5.

Omnibus model of sleep disturbance components as predictors of passive and active suicide ideation.

Passive SI Active SI
Fixed effects Predictor
Est
Posterior SD		 95% CI (lower, upper)
Est
Poster SD
95% CI (lower, upper)
Level 1 Lagged passive SI 0.37 0.12 0.16, 0.64 0.18 0.11 −0.01, 0.43
Lagged active SI 0.21 0.10 0.02,0.43 0.39 0.13 0.16, 0.65
Lagged depressed mood −0.05 0.04 −0.13, 0.04 0.01 0.05 −0.08, 0.11
Days passed −0.01 0.03 −0.07, 0.06 −0.11 0.04 −0.19, −0.04
SOL −0.01 0.03 −0.08, 0.06 0.02 0.04 −0.05, 0.09
WASO 0.01 0.03 −0.05, 0.07 0.06 0.04 −0.02, 0.13
Sleep quality −0.13 0.04 −0.21, −0.05 −0.04 0.04 −0.13, 0.05
Nightmares 0.13 0.03 0.06, 0.19 0.05 0.04 −0.02, 0.12
Level 2 Depressed mood 0.48 0.11 0.26, 0.69 0.28 0.13 0.02, 0.52
SOL 0.11 0.12 −0.12, 0.34 0.26 0.13 0.00, 0.51
WASO 0.07 0.12 −0.17, 0.32 0.09 0.15 −0.20, 0.37
Sleep quality −0.13 0.13 −0.39, 0.12 0.11 0.15 −0.21, 0.38
Nightmares 0.19 0.11 −0.03, 0.41 -.05 0.14 −0.32, 0.21
Residual variance 0.64 0.11 0.35, 0.81 0.76 0.12 0.45, 0.93
Open in a new tab

Note. SI = suicide ideation; SOL = sleep onset latency; WASO = wake after sleep onset; Days passed = time elapsed since the first ecological momentary assessment (EMA) . Level 1 = within-person fluctuations at the daily level. Level 2 = between-persons differences across the assessment period. Standardized point estimates (Est) are reported. Bolded estimates are significant (i.e., 95% CIs do not cross zero).

At the between-persons level, only the effect of longer overall sleep onset latency on the likelihood of active SI during the sampling period remained significant. In contrast, the between-persons effects of nightmares, sleep quality, and sleep onset latency on the likelihood of passive SI during the sampling period were no longer significant when accounting for other sleep disturbance components (see Table 5).

3.4. Exploratory reverse models

Examination of reverse models with daily passive and active SI predicting the subsequent night’s sleep disturbance components, controlling for previous night’s sleep disturbance, revealed that neither passive nor active SI predicted within-person changes in any component of the subsequent night’s sleep (see Table 6).

Table 6.

Passive and active suicide ideation as predictors of sleep disturbance components.

Fixed effects Predictor Est Posterior SD 95% CI (lower, upper)
SOL
Level 1 Lagged passive SI 0.02 0.04 −0.06, 0.10
Lagged active SI −0.05 0.05 −0.14, 0.04
Lagged depressed mood 0.03 0.03 −0.03, 0.08
Days passed −0.11 0.02 −0.16, −0.07
Lagged SOL 0.09 0.03 0.04, 0.14
Residual variance 0.97 0.01 0.95, 0.99
Level 2 Depressed mood 0.10 0.12 −0.14, 0.35
Passive SI 0.19 0.12 −0.06, 0.42
Active SI 0.14 0.13 −0.12, 0.40
Residual variance 0.89 0.05 0.76, 0.97
WASO
Level 1 Lagged passive SI 0.02 0.04 −0.06, 0.10
Lagged active SI −0.02 0.05 −0.12, 0.07
Lagged depressed mood 0.02 0.03 −0.04, 0.07
Lagged WASO 0.13 0.03 0.08, 0.18
Days passed −0.01 0.02 −0.05, 0.04
Residual variance 0.98 0.01 0.96, 0.99
Level 2 Depressed mood 0.22 0.13 −0.04, 0.46
Passive SI 0.21 0.13 −0.05, 0.45
Active SI 0.01 0.14 −0.26, 0.29
Residual variance 0.86 0.06 0.71, 0.96
Sleep Duration
Level 1 Lagged passive SI −0.06 0.04 −0.15, 0.03
Lagged active SI 0.06 0.05 −0.04, 0.16
Lagged depressed mood 0.03 0.03 −0.02, 0.08
Lagged sleep duration <0.01 0.03 −0.05, 0.05
Days passed 0.04 0.02 −0.01, 0.08
Residual variance 0.99 0.01 0.96, 0.99
Level 2 Depressed mood −0.15 0.13 −0.39, 0.11
Passive SI −0.11 0.14 −0.37, 0.17
Active SI 0.11 0.15 −0.21, 0.38
Residual variance 0.91 0.07 0.72, 0.99
Sleep Timing
Level 1 Lagged passive SI −0.01 0.04 −0.09, 0.07
Lagged active SI −0.01 0.05 −0.10, 0.09
Lagged depressed mood −0.06 0.03 −0.11, −0.02
Lagged sleep timing 0.18 0.03 0.13, 0.23
Days passed <0.01 0.02 −0.04, 0.05
Residual variance 0.96 0.01 0.93, 0.98
Level 2 Depressed mood 0.05 0.12 −0.19, 0.29
Passive SI 0.12 0.13 −0.14, 0.35
Active SI −0.07 0.14 −0.33, 0.23
Residual variance 0.94 0.06 0.78, 0.99
Sleep Quality
Level 1 Lagged passive SI −0.03 0.05 −0.11, 0.06
Lagged active SI 0.02 0.05 −0.07, 0.12
Lagged depressed mood −0.02 0.03 −0.07, 0.03
Lagged sleep quality 0.11 0.02 0.07, 0.16
Days passed −0.02 0.02 −0.06, 0.03
Residual variance 0.98 0.01 0.96, 0.99
Level 1 Depressed mood −0.27 0.12 −0.50, −0.03
Passive SI −0.32 0.12 −0.53, −0.07
Active SI 0.18 0.13 −0.09, 0.41
Residual variance 0.76 0.09 0.55, 0.91
Nightmares
Level 1 Lagged passive SI −0.01 0.04 −0.09, 0.06
Lagged active SI 0.01 0.05 −0.08, 0.09
Lagged depressed mood 0.04 0.03 −0.01, 0.09
Lagged nightmares 0.11 0.02 0.07, 0.16
Days passed <−0.01 0.02 −0.05, 0.04
Residual variance 0.98 0.01 0.96, 0.99
Level 2 Depressed mood 0.09 0.13 −0.15, 0.34
Passive SI 0.38 0.12 0.12, 0.58
Active SI −0.20 0.14 −0.44, 0.09
Residual variance 0.78 0.11 0.53, 0.94
Open in a new tab

Note. SI = suicide ideation; SOL = sleep onset latency; WASO = wake after sleep onset; Days passed = time elapsed since the first ecological momentary (EMA) . Level 1 = within-person fluctuations at the daily level. Level 2 = between-persons differences across the assessment period. Standardized point estimates (Est) are reported. Bolded estimates are significant (i.e., 95% CIs do not cross zero).

4. Discussion

The present study used EMA to examine near-term prospective associations between daily changes in sleep disturbance components and next-day likelihood of passive and active SI, adjusting for depressed mood and prior-day SI. These findings disentangle within- and between-person effects and provide evidence for daily intraindividual associations between multiple sleep disturbance components and near-term increases in passive and active SI. This is consistent with the prevention-oriented risk formulation approach (Pisani et al., 2016) which proposes the need to conceptualize the fluid nature of a person’s risk by considering dynamic factors that influence their risk state (within-persons factors) in the context of a more stable risk status (between-person’s factors) in order to directly inform individualized treatment approaches.

This study offers the first evidence for a daily within-person link between nightmares and subsequent near-term increases in passive SI in young adults. This finding is consistent with previous studies finding nightmares predicted next-day SI in adolescents (Glenn et al., 2021) and self-harm thoughts and behaviors more broadly in adults (Hochard et al., 2015), as well as a recent meta-analysis highlighting nightmares as a prospective predictor of SI, attempts, and deaths (Liu et al., 2020). At the within-person level, nights characterized by more nightmares than typical for the individual were followed by days with an increased likelihood of experiencing passive, but not active, SI adjusting for prior day’s SI and depressed mood. This effect remained significant in the omnibus model, highlighting the robustness of this association. In addition, we also found an effect of nightmares on passive SI at the between-person level. That is, participants who reported more nightmares on average were also more likely to report passive SI during the sampling period, though this effect was not a unique predictor over other sleep disturbance components. These findings are consistent with prior work linking nightmares to suicide and self-injury-related outcomes (Andrews and Hanna, 2020) and make two important contributions: (1) nightmares distinguish individuals who are likely to experience passive but not active SI, and (2) nightmares are associated with increased risk for next-day passive SI. If these findings replicate, nightmares may be indicative of more general elevations in distress associated with passive SI but are unlikely to be informative of clinically significant suicide risk or a target for reducing risk for SI.

Sleep quality also predicted passive and active SI. At the within-person level, nights characterized by lower sleep quality than typical for the individual were followed by days with an increased likelihood of experiencing passive and active SI, adjusting for prior day’s SI and depressed mood. At the between-person level, poorer sleep quality on average was associated with a greater likelihood of experiencing passive, but not active, SI during the sampling period. Further, the within-person, but not between-person, effect of sleep quality on passive SI held in the omnibus model, suggesting poor sleep quality does not uniquely distinguish individuals who are likely to experience passive SI from those who do not; however, it does appear to be a unique near-term risk factor for elevated passive ideation among vulnerable individuals. In contrast, the within-person effect of sleep quality on active SI did not remain significant in the omnibus model, suggesting this effect is not unique from other components of sleep disturbance. These findings are consistent with recent work by Littlewood et al. (2019), who observed an effect of sleep quality on next-day SI and extends these findings by distinguishing effects of sleep quality on next-day passive and active SI.

This study is also the first to examine the associations between daily wake after sleep onset and SI in adults. Our findings revealed that nights characterized by more time awake in the night than typical for the individual were followed by days with an increased likelihood of experiencing passive and active SI. However, these within-person effects did not remain significant in the omnibus model, consistent with a recent study in adolescents (Glenn et al., 2021). Further, wake after sleep onset was unrelated to passive or active SI at the between-persons level, suggesting time awake in the night does not uniquely characterize those who experience SI from those who do not. This contrasts with a recent study that showed significant between-persons effects of nocturnal wakefulness, particularly between 11:00 pm and 5:00 am, on SI (Tubbs et al., 2021). Wake after sleep onset may predict near-term increases in SI, whereas wake after sleep onset may only differentiate individuals with SI from those who do not if it occurs during specific times; however, more research is needed to draw firmer conclusions.

We also found an effect of sleep onset latency on passive and active SI at the between-person level, such that those who reported taking longer to fall asleep on average also were more likely to report passive and active SI during the sampling period; sleep onset latency was unrelated to passive or active SI at the within-persons level. These findings are consistent with a previous EMA study in adults with borderline personality disorder, which also found a positive effect of sleep onset latency on SI (composite including passive and active SI with plan and intent) at the between-person, but not within-person level (Kaurin et al., 2021). Likewise, recent studies examining relations between sleep disturbance and SI in large samples of community adults and college students indicate a link between longer sleep onset latency and elevated active SI (Batterham et al., 2021; Khader et al., 2020) and suicide attempts (Batterham et al., 2021). Thus, individuals who experience difficulties falling asleep also tend to report higher rates of passive and active SI. Interestingly, a recent EMA study in adolescents recently discharged from acute care for suicide risk found that longer sleep onset latency was associated with increased next-day SI (Glenn et al., 2021), raising the possibility that a near-term relation between difficulties with sleep initiation and SI may become more enduring if sleep initiation problems are maintained through development.

We found no relation between sleep duration and passive or active SI. Although this finding is inconsistent with one EMA study finding shorter subjective and objective sleep duration predicted next-day SI (Littlewood et al., 2019), it parallels most of the extant EMA and daily diary literature (Glenn et al., 2021; Hamilton et al., 2022; Kaurin et al., 2021). Notably, these findings are discrepant from the non-EMA literature, which has frequently reported a link between short sleep duration and SI and suicidal behaviors (Becker et al., 2018; Dolsen et al., 2021; Gong et al., 2020; Khader et al., 2020). Further, sleep timing was unrelated to passive or active SI in the present study, which is inconsistent with a recent study finding earlier sleep timing predicted increased next-day SI in treatment-seeking adolescents and young adults (Hamilton et al., 2022). Though research examining associations between sleep timing and suicide-related outcomes is limited, one study found a trend-level relation between later bedtimes and increased SI in adults (Rumble et al., 2020), and another study found irregular sleep timing was associated with non-suicidal self-injury among undergraduates (Burke et al., 2022). These discrepant findings may reflect differences in methodology. Retrospective questionnaires may demonstrate bias relative to sleep diaries (Dietch and Taylor, 2021). Indeed, one of the benefits of EMA is reduced retrospective bias (Shiffman et al., 2008). EMA and daily diary approaches may be necessary to delineate relations between dynamic processes that unfold on a day-to-day scale, such as sleep disturbance and suicide risk, and additional EMA research is needed to clarify discrepant findings for the relation between daily sleep duration and SI.

Notably, tests of reverse models found no relation between daily passive or active SI and any component of the subsequent night’s sleep, suggesting a unidirectional relation between sleep disturbance and subsequent SI. Similarly, one prior EMA study found no effect of SI on subsequent subjective or objective sleep (Littlewood et al., 2019), and a longitudinal study found no predictive effect of SI on insomnia symptoms over 1 month (Ribeiro et al., 2012). Together these findings suggest that sleep disturbance temporally precedes SI, rather than sleep disturbance arising as an epiphenomenon or behavioral response to SI and/or suicide-related arousal or negative affect.

These findings can be considered within the context of the “mind after midnight” hypothesis, which suggests that “being awake when reason sleeps” is a period of vulnerability for risky behaviors, such as suicide (Tubbs et al., 2022). Indeed, we found associations between SI and aspects of sleep disturbance that result in excessive nocturnal wakefulness (e.g., sleep onset latency, wake after sleep onset, nightmares) or the perception of nocturnal wakefulness (e.g., poor sleep quality). Should our findings replicate, regular difficulties falling or staying asleep should be included in standard risk assessments, as these sleep difficulties may precede the emergence of passive and active SI providing an opportunity for prevention approaches. Additionally, given rates of non-disclosure to professionals (Calear and Batterham, 2019; McGillivray et al., 2022), patients may be more willing to report sleep difficulties than active SI. Clinicians working with individuals primarily on sleep concerns should consider routinely assessing suicide risk. Additionally, poorer sleep quality and more time awake after sleep onset seem to have more proximal links to active SI and may be acute risk factors for clinically elevated SI among at-risk individuals, whereas longer sleep onset latency may be characteristic of those who tend to experience active SI. From the prevention-oriented risk formulation perspective (Pisani et al., 2016), sleep onset latency may inform risk status, but clinicians should consider assessing for changes in sleep quality and difficulty staying asleep as these dynamic risk factors are indicative of higher risk states that may warrant a different treatment approach for that individual at that point in time.

Given recent findings that cognitive behavior therapy for insomnia reduces SI (Kalmbach et al., 2022), sleep disturbances may have clinical significance for treatment of suicide risk. Behavioral strategies targeting sleep may reduce SI, and encouraging patients to use sleep patterns to indicate when to use their coping skills and safety plan may be beneficial. For example, a night characterized by lower-than-typical sleep quality or higher-than-typical difficulty staying asleep may cue an atrisk individual to be intentional about using their coping skills and safety plan the following day. Should these findings translate to objective sleep, wearable sleep monitoring devices or sleep tracking apps, which are increasingly common (Zambotti et al., 2019), may have clinical utility for providing patients or providers with real-time signals for risk management. Further, nightmares, poorer sleep quality, and longer sleep onset latency appear to be more common among individuals who report passive SI; nightmares, poorer sleep quality, and longer wake after sleep onset are also indicative of acute increases in passive SI within-persons. Although these thoughts do not indicate clinically significant suicide risk in isolation (Chu et al., 2015), they reflect psychiatric distress that may warrant psychiatric care. We agree with Liu et al. (2019) that more longitudinal and experimental research is needed to tease apart which sleep disturbance components are acute risk factors for which suicide-related outcomes (e.g., passive vs active SI, suicide attempts) to make critical advances in risk assessment and intervention.

Although our study adds to the extant literature by highlighting a unidirectional relation between specific sleep disturbance components and next-day passive and active SI, the study limitations must be considered. It is possible there are important, unmeasured mechanisms underlying these associations. Previous findings largely based on cross-sectional data may serve as a guide for potential mechanisms (e.g., executive functioning, social functioning, emotion dysregulation, hopelessness; Hochard et al. 2015, Liu et al. 2020, Littlewood et al. 2016, Russell et al. 2018, Tubbs et al. 2022, Ward-Ciesielski et al. 2018) that could be included in future prospective studies designed to directly test the “mind after midnight” hypothesis. Although we distinguished passive from active SI, we did not examine more severe levels of SI (e.g., planning, suicide intent) and suicidal behaviors (e.g., preparatory behaviors, aborted attempts, suicide attempts) because these were reported infrequently during the study. This research may necessitate higher risk samples and study designs that involve EMA bursts over longer-term follow-up periods to capture suicidal behaviors. Additionally, given that both sleep (Ohayon et al., 2004) and suicide risk changes across the lifespan (CDC, 2023), it is unclear whether these findings generalize across the lifespan, including children, adolescents, and older adults. Further, we did not assess the sample for clinical sleep disorders, and these associations may appear different among individuals with acute versus chronic sleep issues (Liu et al., 2020). We also did not assess SI during nocturnal awakenings, which may have limited our ability to detect the diurnal peak in suicidality. For example, nighttime wakefulness specifically between the hours of 4:00 AM and 4:50 AM was associated with SI among individuals with mood disorders (Ballard et al., 2016). We also focused on several subjective indicators of sleep disturbance, but other subjective aspects of sleep disturbance may further clarify the prospective associations between sleep disturbance and suicide risk, such as nightmare distress or intensity (Hochard et al., 2019; Lee and Suh, 2016). Moreover, it is possible that examining multiple components of sleep disturbance contributes to the heterogeneity observed in the extant literature on sleep disturbance and suicide. Although we chose a robust statistical approach to address other potential sources of inconsistencies and improve our understanding of the short-term impact of individual sleep disturbance components, future researchers should consider alternative approaches (e.g., latent factors, composite variables) that could synthesize findings across studies. Lastly, our findings are limited by exclusive reliance on subjective sleep assessment. Given the known discrepancies between subjective and objective sleep measures (Dietch and Taylor, 2021; Harvey and Tang, 2012), integrating these measures may also be necessary to understand the complex associations between sleep disturbance and suicide risk. Despite these limitations, this study offers novel insight into the role of daily sleep disturbance as a near-term predictor of SI and highlights the need for additional research on sleep disturbance and suicide-related outcomes.

Role of the funding source

This research was supported, in part, by grants from the National Institute of Mental Health (R01 MH115922; T32 MH018269) and the National Heart, Lung, and Blood Institute (T32 HL149646). The funders had no involvement in study design, in the collection, analysis and interpretation of data, in the writing of the report, or in the decision to submit the article for publication.

Footnotes

CRediT authorship contribution statement

Rebecca C. Cox : Conceptualization, Formal analysis, Data curation, Writing – review & editing. Sarah L. Brown : Conceptualization, Formal analysis, Data curation, Investigation, Writing – review & editing. Brittany N. Chalmers : Conceptualization, Investigation, Writing – review & editing. Lori N. Scott : Conceptualization, Formal analysis, Resources, Writing – review & editing, Supervision, Project administration, Funding acquisition.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

1

Given interpretability issues that may result from adjusting for depression (Mitchell et al., 2017; Rogers et al., 2018), we also tested the hypothesized models without depressed mood as a covariate. The pattern and significance of the results were highly consistent. The only exception was the emergence of positive between-persons effects of sleep onset latency on active suicide ideation and wake after sleep onset on passive suicide ideation, consistent with zero-order correlations, which became non-significant in the omnibus model. There-fore, models adjusting for depressed mood were retained.

References

  1. Andrews S, Hanna P, 2020. Investigating the psychological mechanisms underlying the relationship between nightmares, suicide and self-harm. Sleep Med. Rev 54, 101352. 10.1016/j.smrv.2020.101352. [DOI] [PubMed] [Google Scholar]
  2. Ballard ED, Vande Voort JL, Bernert RA, Luckenbaugh DA, Richards EM, Niciu MJ, Furey ML, Duncan WC Jr, Zarate CA Jr, 2016. Nocturnal wakefulness is associated with next-day suicidal ideation in major depression and bipolar disorder. J. Clin. Psychiatry 77, 825–831. 10.4088/JCP.15m09943.Nocturnal. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Batterham PJ, Werner-Seidler A, Calear AL, McCallum S, Gulliver A, 2021. Specific aspects of sleep disturbance associated with suicidal thoughts and attempts. J. Affect. Disord 282, 574–579. 10.1016/j.jad.2020.12.150. [DOI] [PubMed] [Google Scholar]
  4. Becker SP, Dvorsky MR, Holdaway AS, Luebbe AM, 2018. Sleep problems and suicidal behaviors in college students. J. Psychiatr. Res 99, 122–128. 10.1016/j.jpsychires.2018.01.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bishop TM, Walsh PG, Ashrafioun L, Lavigne JE, Pigeon WR, 2020. Sleep, suicide behaviors, and the protective role of sleep medicine. Sleep Med 66, 264–270. 10.1016/j.sleep.2019.07.016. [DOI] [PubMed] [Google Scholar]
  6. Brown GK, Jager-Hyman S, 2014. Evidence-based psychotherapies for suicide prevention. Am. J. Prev. Med 47, S186–S194. 10.1016/j.amepre.2014.06.008. [DOI] [PubMed] [Google Scholar]
  7. Burke TA, Hamilton JL, Seigel D, Kautz M, Liu RT, Alloy LB, Barker DH, 2022. Sleep irregularity and nonsuicidal self-injurious urges and behaviors. Sleep zsac084 10.1093/sleep/zsac084. [DOI] [PMC free article] [PubMed]
  8. Calear AL, Batterham PJ, 2019. Suicidal ideation disclosure: patterns, correlates and outcome. Psychiatry Res 278, 1–6. 10.1016/j.psychres.2019.05.024. [DOI] [PubMed] [Google Scholar]
  9. Carney CE, Buysse DJ, Ancoli-Israel S, Edinger JD, Krystal AD, Lichstein KL, Morin CM, 2012. The consensus sleep diary: standardizing prospective sleep self-monitoring. Sleep 35, 287–302. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Chan NY, Zhang J, Tsang CC, Li AM, Chan JWY, Wing YK, Li SX, 2020. The associations of insomnia symptoms and chronotype with daytime sleepiness, mood symptoms and suicide risk in adolescents. Sleep Med 74, 124–131. 10.1016/j.sleep.2020.05.035. [DOI] [PubMed] [Google Scholar]
  11. Chu C, Klein KM, Buchman-Schmitt JM, Hom MA, Hagan CR, Joiner TE, 2015. Routinized assessment of suicide risk in clinical practice: an empirically informed update. J. Clin. Psychol 71, 1186–1200. 10.1002/jclp.22210. [DOI] [PubMed] [Google Scholar]
  12. Coppersmith DDL, Kleiman EM, Glenn CR, Millner AJ, Nock MK, 2019. The dynamics of social support among suicide attempters: a smartphone-based daily diary study. Behav. Res. Ther 120, 103348. 10.1016/j.brat.2018.11.016. [DOI] [PubMed] [Google Scholar]
  13. Dietch JR, Taylor DJ, 2021. Evaluation of the Consensus Sleep Diary in a community sample: comparison with single-channel EEG, actigraphy, and retrospective questionnaire. J. Clin. Sleep Med 17, 1389–1399. 10.5664/jcsm.9200. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Dolsen MR, Prather AA, Lamers F, Penninx BWJH, 2021. Suicidal ideation and suicide attempts: associations with sleep duration, insomnia, and inflammation. Psychol. Med 51, 2094–2103. 10.1017/S0033291720000860. [DOI] [PubMed] [Google Scholar]
  15. First MB, Williams JBW, Karg RS, Spitzer RL, 2015. Structured Clinical Interview For DSM-5-Research Version American Psychiatric Association, Arlington, VA. [Google Scholar]
  16. Fox KR, Huang X, Guzmán EM, Funsch KM, Cha CB, Ribeiro JD, Franklin JC, 2020. Interventions for suicide and self-injury: a meta-analysis of randomized controlled trials across nearly 50 years of research. Psychol. Bull 146, 1117–1145. 10.1037/bul0000305. [DOI] [PubMed] [Google Scholar]
  17. Franklin JC, Ribeiro JD, Fox KR, Bentley KH, Kleiman EM, Huang X, Musacchio KM, Jaroszewski AC, Chang BP, Nock MK, 2017. Risk factors for suicidal thoughts and behaviors: a meta-analysis of 50 years of research. Psychol. Bull 143, 187–232. 10.1037/bul0000084. [DOI] [PubMed] [Google Scholar]
  18. Glenn CR, Kleiman EM, Kearns JC, Boatman AE, Conwell Y, Alpert-Gillis LJ, Pigeon W, 2021. Sleep problems predict next-day suicidal thinking among adolescents: a multimodal real-time monitoring study following discharge from acute psychiatric care. Dev. Psychopathol 33, 1701–1721. 10.1017/S0954579421000699. [DOI] [Google Scholar]
  19. Gong Q, Li S, Wang S, Li H, Han L, 2020. Sleep and suicidality in school-aged adolescents: a prospective study with 2-year follow-up. Psychiatry Res 287, 112918. 10.1016/j.psychres.2020.112918. [DOI] [PubMed] [Google Scholar]
  20. Gratch I, Choo TH, Galfalvy H, Keilp JG, Itzhaky L, Mann JJ, Oquendo MA, Stanley B, 2021. Detecting suicidal thoughts: the power of ecological momentary assessment. Depress. Anxiety 38, 8–16. 10.1002/da.23043. [DOI] [PubMed] [Google Scholar]
  21. Hallensleben N, Spangenberg L, Forkmann T, Rath D, Hegerl U, Kersting A, Kallert TW, Glaesmer H, 2018. Investigating the dynamics of suicidal ideation: preliminary findings from a study using ecological momentary assessments in psychiatric inpatients. Crisis 39, 65–69. 10.1027/0227-5910/a000464. [DOI] [PubMed] [Google Scholar]
  22. Hamilton JL, Tsypes A, Zelazny J, Sewall CJR, Rode N, Merranko J, Brent DA, Goldstein TR, Franzen PL, 2022. Sleep influences daily suicidal ideation through affective reactivity to interpersonal events among high-risk adolescents and young adults. J. Child Psychol. Psychiatry 10.1111/jcpp.13651. [DOI] [PMC free article] [PubMed]
  23. Harvey AG, Tang N, 2012. Mis)perception of sleep in insomnia: a puzzle and a resolution. Psychol. Bull 138, 77–101. 10.1037/a0025730. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Hedström AK, Hössjer O, Bellocco R, Ye W, Trolle LY, Åkerstedt T, 2021. Insomnia in the context of short sleep increases suicide risk. Sleep 44, 1–7. 10.1093/sleep/zsaa245. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Hochard KD, Ashcroft S, Carroll J, Heym N, Townsend E, 2019. Exploring thematic nightmare content and associated self-harm risk. Suicide Life Threat. Behav 49, 64–75. 10.1111/sltb.12402. [DOI] [PubMed] [Google Scholar]
  26. Hochard KD, Heym N, Townsend E, 2015. The unidirectional relationship of nightmares on self-harmful thoughts and behaviors. Dreaming 25, 44–58. 10.1037/a0038617. [DOI] [Google Scholar]
  27. Jahn DR, Muralidharan A, Drapalski AL, Brown CH, Fang J, Lucksted A, 2018. Differences in suicide and death ideation among veterans and non-veterans with serious mental illness. Psychol. Serv 15, 31–39. 10.1037/ser0000127. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Kalmbach DA, Cheng P, Ahmedani BK, Peterson EL, Reffi AN, Sagong C, Seymour GM, Ruprich MK, Drake CL, 2022. Cognitive-behavioral therapy for insomnia prevents and alleviates suicidal ideation: insomnia remission is a suicidolytic mechanism. Sleep 45. 10.1093/sleep/zsac251. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Kaurin A, Hisler G, Dombrovski AY, Hallquist MN, Wright AGC, 2021. Sleep and next-day negative affect and suicidal ideation in borderline personality disorder. Pers. Disord. Theory Res. Treat 10.1037/per0000496. [DOI] [PMC free article] [PubMed]
  30. Khader S, Tubbs AS, Haghighi A, Athey A, Killgore WDS, Hale L, Branas C, Gehrels JA, Alfonso-Miller P, Fernandez FX, Grandner MA, 2020. Onset insomnia and insufficient sleep duration are associated with suicide ideation in university students and athletes. J. Affect. Disord 274, 1161–1164. 10.1016/j.jad.2020.05.102.Onset. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Kleiman EM, Turner BJ, Fedor S, Beale EE, Huffman JC, Nock MK, 2017. Examination of real-time fluctuations in suicidal ideation and its risk factors: results from two ecological momentary assessment studies. J. Abnorm. Psychol 126, 726–738. 10.1037/abn0000273. [DOI] [PubMed] [Google Scholar]
  32. Klonsky ED, May AM, 2015. The Three-Step Theory (3ST): a new theory of suicide rooted in the “Ideation-to-Action” framework. Int. J. Cogn. Ther 8, 114–129. [Google Scholar]
  33. Lee R, Suh S, 2016. Nightmare distress as a mediator between nightmare frequency and suicidal ideation. Dreaming 26, 308–318. 10.1037/drm0000039. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Littlewood DL, Kyle SD, Carter LA, Peters S, Pratt D, Gooding P, 2019. Short sleep duration and poor sleep quality predict next-day suicidal ideation: an ecological momentary assessment study. Psychol. Med 49, 403–411. 10.1017/S0033291718001009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Liu J, Tu Y, Lai Y, Lee H, Tsai P, Chen T, Huang H, Chen Y, Chiu H, 2019. Associations between sleep disturbances and suicidal ideation, plans, and attempts in adolescents: a systematic review and meta-analysis. Sleep 42, zsz054. 10.1093/sleep/zsz054. [DOI] [PubMed] [Google Scholar]
  36. Liu RT, Steele SJ, Hamilton JL, Do QBP, Furbish K, Burke TA, Martinez AP, Gerlus N, 2020. Sleep and suicide: a systematic review and meta-analysis of longitudinal studies. Clin. Psychol. Rev 81, 101895. 10.1016/j.cpr.2020.101895. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Liu X, Yang Y, Liu ZZ, Jia CX, 2021. Longitudinal associations of nightmare frequency and nightmare distress with suicidal behavior in adolescents: mediating role of depressive symptoms. Sleep 44, 1–9. 10.1093/sleep/zsaa130. [DOI] [PubMed] [Google Scholar]
  38. McGillivray L, Rheinberger D, Wang J, Burnett A, Torok M, 2022. Non-disclosing youth: a cross sectional study to understand why young people do not disclose suicidal thoughts to their mental health professional. BMC Psychiatry 22. 10.1186/s12888-021-03636-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. McNeish D, 2016. Estimation methods for mixed logistic models with few clusters estimation methods for mixed logistic models with few clusters. Multivariate Behav. Res 51, 790–804. 10.1080/00273171.2016.1236237. [DOI] [PubMed] [Google Scholar]
  40. Millner AJ, Lee MD, Nock MK, 2017. Describing and measuring the pathway to suicide attempts: a preliminary study. Suicide Life Threat. Behav 47, 353–369. 10.1111/sltb.12284. [DOI] [PubMed] [Google Scholar]
  41. Monk TH, Reynolds CF, Kupfer DJ, Buysse DJ, Coble PA, Hayes AJ, Machen MA, Petrie SR, Ritenour AM, 1994. The Pittsburgh sleep diary. J. Sleep Res 3, 111–120. [PubMed] [Google Scholar]
  42. Nardorff MR, Nazem S, Fiske A, 2013. Insomnia symptoms, nightmares, and suicide risk: duration of sleep disturbance matters. Suicide Life Threat. Behav 43, 139–149. 10.1111/sltb.12003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Nock MK, Prinstein MJ, Sterba SK, 2009. Revealing the form and function of self-injurious thoughts and behaviors: a real-time ecological assessment study among adolescents and young adults. J. Abnorm. Psychol 118, 816–827. 10.1037/a0016948. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Ohayon MM, Carskadon MA, Guilleminault C, Vitiello MV, 2004. Meta-analysis of quantitative sleep parameters from childhood to old age in healthy individuals: developing normative sleep values across the human lifespan. Sleep 27, 1255–1273. 10.1080/16506073.2015.1026386. [DOI] [PubMed] [Google Scholar]
  45. O’Riley AA, Van Orden KA, He H, Richardson TM, Podgorski C, Conwell Y, 2014. Suicide and death ideation in older adults obtaining aging services. Am. J. Geriatr. Psychiatry 22, 614–622. 10.1016/j.jagp.2012.12.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Pfohl B, Blum N, Zimmerman M, 1997. Structured Clinical Interview For DSM-IV Personality (SIDP-IV) American Psychiatric Press, Washington DC. [Google Scholar]
  47. Pigeon WR, Pinquart M, Conner K, 2012. Meta-analysis of sleep disturbance and suicidal thoughts and behaviors. J. Clin. Psychiatry 73, 1160–1167. 10.4088/JCP.11r07586. [DOI] [PubMed] [Google Scholar]
  48. Pisani AR, Murrie DC, Silverman MM, 2016. Reformulating suicide risk formulation: from prediction to prevention. Acad. Psychiatry 40, 623–629. 10.1007/s40596-015-0434-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Posner K, Brown GK, Stanley B, Brent DA, Yershova KV, Oquendo MA, Currier GW, Melvin GA, Greenhill L, Shen S, Mann JJ, 2011. The Columbia-suicide severity rating scale: internal validity and internal consistency findngs from three multisite studies with adolescents and adults. Am.J. Psychiatry 168, 1266–1277. 10.1176/appi.ajp.2011.10111704. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Ribeiro JD, Pease JL, Gutierrez PM, Silva C, Bernert RA, Rudd MD, Joiner TE, 2012. Sleep problems outperform depression and hopelessness as cross-sectional and longitudinal predictors of suicidal ideation and behavior in young adults in the military. J. Affect. Disord 136, 743–750. 10.1016/j.jad.2011.09.049. [DOI] [PubMed] [Google Scholar]
  51. Riblet NBV, Shiner B, Young-Xu Y, Watts BV, 2017. Strategies to prevent death by suicide: meta-analysis of randomised controlled trials. Br. J. Psychiatry 210, 396–402. 10.1192/bjp.bp.116.187799. [DOI] [PubMed] [Google Scholar]
  52. Rumble ME, McCall WV, Dickson DA, Krystal AD, Rosenquist PB, Benca RM, 2020. An exploratory analysis of the association of circadian rhythm dysregulation and insomnia with suicidal ideation over the course of treatment in individuals with depression, insomnia, and suicidal ideation. J. Clin. Sleep Med 16, 1311–1319. 10.5664/jcsm.8508. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Russell K, Rasmussen S, Hunter SC, 2018. Insomnia and nightmares as markers of risk for suicidal ideation in young people: investigating the role of defeat and entrapment. J. Clin. Sleep Med 14, 775–784. 10.5664/jcsm.7104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Sedano-Capdevila A, Porras-Segovia A, Bello HJ, Baca-García E, Barrigon ML, 2021. Use of ecological momentary assessment to study suicidal thoughts and behavior: a systematic review. Curr. Psychiatry Rep 23, 41. 10.1007/s11920-021-01255-7. [DOI] [PubMed] [Google Scholar]
  55. Shiffman S, Stone AA, Hufford MR, 2008. Ecological momentary assessment. Annu. Rev. Clin. Psychol 4, 1–32. 10.1002/9781118384404.ch20. [DOI] [PubMed] [Google Scholar]
  56. Tubbs AS, Fernandez F, Perlis ML, Hale L, Branas CC, Barrett M, Chakravorty S, Khader W, Grandner MA, 2021. Suicidal ideation is associated with nighttime wakefulness in a community sample. Sleep 44, 1–9. 10.1093/sleep/zsaa128. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Tubbs AS, Fernandez FX, Grandner MA, Perlis ML, Klerman EB, 2022. The mind after midnight: nocturnal wakefulness, behavioral dysregulation, and psychopathology. Front. Netw. Physiol 10.3389/fnetp.2021.830338.The. [DOI] [PMC free article] [PubMed]
  58. Van Orden KA, Witte TK, Cukrowicz KC, Braithwaite S, Selby EA, Joiner TE, 2010. The interpersonal theory of suicide. Psychol. Rev 117, 575–600. 10.1037/a0018697. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Zambotti M, Cellini N, Goldstone A, Colrain IM, Baker FC, 2019. Wearable sleep technology in clinical and research settings. Med. Sci. Sports Exerc 51, 1538–1557. 10.1249/MSS.0000000000001947. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES