0656 Demographic and Clinical Features of Nocturnal Suicide

Introduction

The risk for suicide is greatest at night after adjusting for population wakefulness, possibly due to sleep- and circadian-dependent changes in neurophysiology to promote sleep. Those who die by suicide at night, however, may differ by demographic and/or clinical characteristics from those who die by suicide during the day.

Methods

An archival analysis of the National Violent Death Reporting System for 2003-2017 identified 77,784 suicide deaths with time of fatal injury. Cases were divided into daytime (5AM to 10:59PM) or nighttime (11PM to 4:59AM) and characterized by age, sex, race, ethnicity, marital status, military service, education, prior diagnosis of an anxiety disorder, bipolar disorder, depression, history of suicidal ideation, PTSD, and schizophrenia, as well as the presence of an opiate or cannabis, and blood alcohol level (BAL) on autopsy. Bidirectional stepwise regression and robust Poisson models characterized significant predictors of nocturnal suicide using incident risk ratios (IRR).

Results

Nocturnal and daytime suicides differed on all sociodemographic variables. Nocturnal suicides were more prevalent among those with bipolar disorder, PTSD, an elevated BAL, and those who tested positive for cannabis. Stepwise models identified a significant age by BAL interaction. Using adults 35-64 with BAL=0mg/dl as the reference, adults 35-64 with a BAL<80mg/dl had a 46% greater risk of suicide at night, and those with a BAL≥80mg/dl had a 78% greater risk. Individuals 15-34 had a nighttime suicide that was 26% greater with BAL=0mg/dl, 84% greater with BAL<80mg/dl, and 298% greater with BAL≥80mg/dl. Conversely, individuals 65 and older were 27% less likely to die at night with BAL=0mg/dl, while those with a BAL>0mg/dl did not differ from those aged 35-64 with BAL=0mg/dl. The risk of nocturnal suicide was also 17% greater among those with a prior history of suicidal ideation, and 13% less likely among those with documented depression.

Conclusion

Nocturnal suicide is more prevalent among intoxicated younger adults and those with previous suicidal ideation. However, suicide victims with depression were less likely to die at night. Further research is needed to target suicide prevention efforts at appropriately times for those with mood, substance, and alcohol use disorders.

Support (If Any)

 

0017 Murder on the Midnight Express: Nocturnal Wakefulness and Homicide Risk

Introduction

There is a nocturnal peak in incident suicide risk after adjusting for population wakefulness (Perlis et al., 2016; Tubbs et al., 2020). This peak in risk is hypothesized to result from a series of negative changes in mood, reward processing, and executive function that occur at night and increase the propensity for dysregulated and violent behaviors. Although the unadjusted incidence of dying by homicide is elevated at night, no existing studies of time-of-day and death by violent crime have adjusted for population wakefulness.

Methods

Data from 48,486 homicide victims with a known time of fatal injury were collected from the National Violent Death Reporting System (NVDRS) for 2003-2017, tabulated by clock hour, age, sex, race, and ethnicity, and combined with population wakefulness data from the American Time Use Survey (ATUS) for the same years. Homicide counts were additionally characterized by the proportion of cases with blood alcohol level (BAL) of 0, <80mg/dl, or ≥80mg/dl at autopsy and modeled using robust Poisson regression with population wakefulness entered as an offset term, thus producing hourly incident risk ratios (IRR).

Results

Homicide counts were lowest in the morning (6AM-7AM) and highest at night (10PM-11PM). After adjusting for population wakefulness, the incident risk for death by homicide was elevated between 10PM and 5AM compared to the 24-hour average, with the highest risk between 2AM (IRR: 8.25 [6.62-10.3]) and 3AM (IRR: 7.22 [6.04-8.64]). Moreover, the adjusted risk of dying by homicide was significantly greater at night for those with a BAL≥80mg/dl, such that the risk at 2AM was 13.8-fold greater than the 24-hour average (IRR: 13.8 [10.6-18.1]).

Conclusion

The risk of homicide death is higher at night after adjusting for population wakefulness and especially among those with alcohol intoxication. Although homicide victims do not choose when to die (unlike suicide victims), neurophysiological changes at night may promote risky behaviors or put victims in more dangerous circumstances than they would be otherwise. Future research should examine sociodemographic, clinical, and circadian risk factors for death by homicide, as well as examine time-of-day patterns in other violent crimes.

Support (If Any)

 

0664 Temporal Patterns of Suicidal Ideation in the Emergency Department

Introduction

Nocturnal wakefulness may be dangerous for vulnerable populations: the incident risk for suicide is highest at night after adjusting for population wakefulness, and nocturnal wakefulness is associated with suicidal ideation. These observations support the hypothesis that sleep- and circadian-dependent changes in mood, reward processing, and executive function increase the risk for disinhibited behavior at night (during periods of nocturnal wakefulness). The present study evaluated this hypothesis by using the timing of emergency department encounters for suicidal ideation.

Methods

An archival analysis of data from two emergency departments (EDs) in Tucson, Arizona from 2018 and 2019 yielded 51,370 encounters for any reason across 29,359 individuals with usable data, and the time of initial contact was extracted for each case. Of these, 571 individuals (1.94%) sought care for 763 (1.49%) instances of suicidal ideation (determined by ICD-10 code R45.861). Encounters were characterized by date/time, age, sex, race/ethnicity, blood alcohol level (if tested), homelessness, and prior diagnosis of a psychotic disorder, bipolar disorder, or depressive disorder. Suicidal ideation encounters were analyzed as raw counts and as a proportion of all encounters by clock hour and time-of-day categories (night: 12AM-5:59AM; morning: 6AM-11:59AM; afternoon:12PM-5:59PM; evening: 6PM-11:59PM) using robust Poisson models.

Results

Although most ED encounters occurred between 6PM and midnight (mean: 9:42PM), the greatest number of suicidal ideation encounters occurred between 12AM and 3AM (mean: 12:18AM). After adjusting for the per-hour proportion of ED visits, the incident risk for a suicidal ideation encounter increased between 8AM and 11AM, peaked at 10AM (IRR: 1.95 [1.10-3.44]) and was lowest at 4PM (IRR: 0.54 [0.32-0.91]). Compared to the evening, the incident risk of suicidal ideation was 64% greater in the morning (IRR: 1.64 [1.31-2.06]), 25% greater at night (IRR: 1.25 [1.00-1.56]), but not different for afternoon encounters.

Conclusion

After adjusting for overall encounter rates, ED encounters for suicidal ideation are more likely to occur in the morning. Although the morning peak in incident risk is later than the reported nocturnal risk for incident suicide, this may reflect a delay between when an individual develops suicidal ideation and when they seek or receive treatment.

Support (If Any)

 

0036 Population Wakefulness and Nocturnal Suicide Risk

Introduction

Nocturnal wakefulness may mediate the relationship between disrupted sleep and suicide risk since nighttime is associated with a peak in negative mood and altered reward processing and executive function. One example is a wakefulness-adjusted nocturnal peak in population suicide risk measured from 2003-2010 (Perlis et al, 2016), but these results have not been replicated with more recent data.

Methods

A total of 77,784 suicides with known time of fatal injury were extracted from the National Violent Death Reporting System (NVDRS) for 2003-2010 and 2011-2017. These data were then weighted by the estimated proportion of the population that was awake at each hour as derived from the American Time Use Survey (ATUS) for the same years. Suicides were tabulated by clock hours, age, sex, race, and ethnicity, and suicide counts were modeled using robust Poisson regression with hourly population wakefulness entered as an offset term, thus producing hourly incident risk ratios.

Results

A comparison of analyses between previously reported data (2003 to 2010) and new data (2011 to 2017) showed a consistently elevated risk of suicide at night (midnight to 6AM) that did not differ between time periods. After combining all fifteen years, the maximum number of suicides occurred at noon. Adjusting for population wakefulness, however, revealed a significant increased risk for suicide between 11PM and 5AM, with a 4.61-fold increase at 3AM (IRR: 4.61 [4.11-5.16]). Adjusting for age, sex, race, and ethnicity attenuated, but did not alter these results. In subgroup analyses, the nocturnal risk for suicide was elevated among those with bipolar disorder (5.2% of cases), those with a blood alcohol level greater than 80 mg/dl (14.9% of cases), and those who tested positive for a Z-drug (i.e., zolpidem, zaleplon, and eszopiclone) at autopsy (0.7% of cases).

Conclusion

Fifteen years of data from suicides across the United States show a significantly increased risk for suicide during the circadian night that peaks at 3AM. Nocturnal wakefulness remains a significant risk factor for suicide, and suicide prevention efforts may benefit from interventions to reduce nocturnal wakefulness and/or an increase in prevention resources at this time.

Support (If Any)

 

0209 The Effect of Time of Day of COVID-19 Vaccination and other Covariates on Side Effects

Introduction

Circadian rhythms have critical roles in human health. We quantified the effect of time-of-day of COVID-19 vaccination and other covariates on self-reported side effects post vaccination.

Methods

The dataset was created from MassGeneralBrigham (MGB) electronic health records and REDCap survey that collected self-reported symptoms for 1-3 days after each immunization. Variables are demographics (age, sex, race, and ethnicity), vaccine manufacturer, clock time of vaccine administration/appointment, any COVID-19 diagnosis/positive test prior to vaccination, any history of allergy, and any note of epinephrine self-injection (e.g., EpiPen) medication. Time of day groupings were morning (6 am–10 am), midday (10 am–2 pm), late afternoon (2 pm–6 pm) or evening (6 pm–10 pm). Side effects were classified as Allergic (Rash; Hives; Swollen lips, tongue, eyes, or face; Wheezing) and Non-Allergic (New Headache, New Fatigue, Arthralgias, Myalgias, Fever) symptoms. The study was approved by the MGB IRB.Machine learning (ML) techniques (e.g., extreme gradient boosting) were applied to the variables to predict the occurrence of side effects. Stratified k-fold cross validation was used to validate the performance of the ML models. Shapley Additive Explanation values were computed to explain the contribution of each of the variables to the prediction of the occurrence of side effects.

Results

Data were from 54,844 individuals. On day 1 after the first vaccination, (i) females, people who received the Moderna vaccine, and those with any allergy history were more likely to report Allergic side effects; and (ii) females, people who received the Janssen vaccine, those who had prior COVID-19 diagnosis ,and those who received their vaccine in the morning or midday and were more likely to report Non-Allergic symptoms. Older persons had fewer side effects of any type.

Conclusion

ML techniques identified demographic and time-of-day-of-vaccination effects on side effects reported on the first day after the first dose of a COVID-19 vaccination. We will use these techniques to test for changes on days 2 and 3 after the first dose, and the first 3 days after the second dose and for the influence of recent night or shiftwork. Future work should target underlying physiological reasons.

Support (If Any)

 

The CLASS Study (Circadian Light in Adolescence, Sleep and School): protocol for a prospective, longitudinal cohort to assess sleep, light, circadian timing and academic performance in adolescence

BACKGROUND

During adolescence, sleep and circadian timing shift later, contributing to restricted sleep duration and irregular sleep-wake patterns. The association of these developmental changes in sleep and circadian timing with cognitive functioning, and consequently academic outcomes, has not been examined prospectively. The role of ambient light exposure in these developmental changes is also not well understood. Here, we describe the protocol for the Circadian Light in Adolescence, Sleep and School (CLASS) Study that will use a longitudinal design to examine the associations of sleep-wake timing, circadian timing and light exposure with academic performance and sleepiness during a critical stage of development. We also describe protocol adaptations to enable remote data collection when required during the COVID-19 pandemic.

METHODS

Approximately 220 healthy adolescents aged 12-13 years (school Year 7) will be recruited from the general community in Melbourne, Australia. Participants will be monitored at five 6 monthly time points over 2 years. Sleep and light exposure will be assessed for 2 weeks during the school term, every 6 months, along with self-report questionnaires of daytime sleepiness. Circadian phase will be measured via dim light melatonin onset once each year. Academic performance will be measured via national standardised testing (National Assessment Program-Literacy and Numeracy) and the Wechsler Individual Achievement Test-Australian and New Zealand Standardised Third Edition in school Years 7 and 9. Secondary outcomes, including symptoms of depression, anxiety and sleep disorders, will be measured via questionnaires.

DISCUSSION

The CLASS Study will enable a comprehensive longitudinal assessment of changes in sleep-wake timing, circadian phase, light exposure and academic performance across a key developmental stage in adolescence. Findings may inform policies and intervention strategies for secondary school-aged adolescents.

ETHICS AND DISSEMINATION

Ethical approval was obtained by the Monash University Human Research Ethics Committee and the Victorian Department of Education. Dissemination plans include scientific publications, scientific conferences, via stakeholders including schools and media.

STUDY DATES

Recruitment occurred between October 2019 and September 2021, data collection from 2019 to 2023.

078 Chronic sleep and circadian disruption differentially affects blood pressure, renal sodium retention, and aldosterone secretion

Introduction

Chronic sleep restriction (CSR) and recurrent circadian disruption (RCD; e.g., rotating shiftwork) can increase an individual’s risk of cardiovascular and kidney disease. However, no study has assessed whether CSR and RCD together increase blood pressure (BP) and alter renal function (RF). We tested the hypotheses that the combination of CSR and RCD would increase blood pressure, renal sodium retention, and aldosterone secretion in individuals living for 3 weeks on an imposed non-24-h sleep-wake (SW) schedule (induces RCD) and controlled diet with or without CSR.

Methods

Seventeen (9M) healthy participants (aged 26.1±4.5y [mean±SD]) were scheduled to twenty-four 20-h Forced Desynchrony days and were randomized to either Control (1:2 sleep:wake, 6.67h sleep:13.33h wake; n=8) or CSR (1:3.3 sleep:wake, 4.67h sleep: 5.33h wake; n=9) SW conditions during a 32-day inpatient protocol. BP was measured following ~80–90 min in constant seated posture after scheduled waketime. All urine voids were collected, combined and sampled in 3-6h blocks throughout the study. Samples were assayed for sodium, potassium and aldosterone and analyzed as both excretion rates and total secretion (both per 20h). Data were assigned circadian phase using fitted core body temperature and analyzed using mixed-effects models with circadian phase, aligned/misaligned sleep, or time awake (with associated scheduled activity, sleep/wake, and feeding behaviors) and their interactions as fixed effects.

Results

There was a significant interaction between aligned/misaligned sleep and condition for resting BP (p=0.02), such that systolic BP was ~6% higher following circadian-misaligned sleep in CSR compared to Control (p=0.04). Renal sodium and potassium followed a robust circadian pattern (p<0.0001), with limited influence of time awake. In contrast, the timing of aldosterone excretion was affected by time awake (p<0.05). Total daily renal sodium secretion decreased from beginning to end of the protocol (p=0.03), with no change in sodium consumption and aldosterone secretion (p=0.95).

Conclusion

Under conditions similar to rotating shiftwork, systolic BP increased and sodium, potassium, and aldosterone were differentially influenced by circadian phase and scheduled behaviors. Additionally, renal sodium secretion decreased despite minimal changes in aldosterone secretion, suggesting increased renal aldosterone sensitivity. These findings may provide insight into mechanisms contributing to poor cardiovascular and renal health observed in shiftwork.

Support (if any):

077 Human activity levels reflect circadian influences independent of sleep/wake behavior

Introduction

Actigraphy is a non-invasive method that allows long-term recordings of activity, light, and other variables in diverse environments. In real-world settings, activity usually has a 24-hour rhythm that may arise from sleep/wake-associated behavior and/or circadian rhythmicity. We tested for an independent circadian component using data from people living on non-24 hours “days” in the laboratory.

Methods

Data are from five inpatient studies with tightly-controlled forced desynchrony (FD) conditions. Participants (19–34 yo) were healthy by history, physical exam, laboratory tests of blood and urine, and clinical polysomnography, and did not report using prescription medicines. Caffeine-containing substances were prohibited during the study. Protocol 1: 7 participants (3 F) T-cycle (i.e., FD sleep-wake cycle duration) = 42.85h; Rest:Activity ratio 1:3.3. Protocol 2: 8 participants (3 F) T cycle =42.85h; Rest:Activity 1:2. Protocol 3: 9 participants (3 F) T cycle =28.0h; Rest:Activity 1:2. Protocol 4: 7 participants (3 F) T cycle =20.0h; Rest:Activity ratio 1:3.3. Protocol 5: 7 participants (5 F) T cycle =20.0h; Rest:Activity 1:2. At all times except during showers, participants wore an actiwatch that measured activity levels and light. Melatonin period and phase 0 (i.e., fit maximum) were computed using non-orthogonal spectral analyses. Data were analyzed relative to 3-hr Circadian Phase bins (1/8 of computed circadian period for each individual) and 3-hr Wake Duration bins. Activity data were summarized using Zero-Inflated-Poison-based statistics for each Circadian*Wake Duration bin for each individual and then across individuals within each study. Repeated measures ANOVA were conducted. Statistics were performed using SAS.

Results

For all protocols, there were significant differences (all p<0.007) by individual participant, by Circadian Phase, and by Wake Duration bin, but not by the interaction term (Circadian Phase* Wake Duration). Highest levels of activity were at Circadian Phase 7.5–10.5 (~10am–1pm) and lowest values at Circadian Phase -1.5–1.5 (~midnight–3 am). Activity values were lowest at scheduled sleep times.

Conclusion

Circadian rhythms independent of sleep/wake behaviors influence activity levels and may be an important component of analyses. In individuals living on non-24-hr days (e.g., some blind people and some sighted people with Non-24-hr Sleep Disorder), it may be possible to derive circadian-based metrics.

Support (if any)

NIH K24-HL105664, P01-AG009975, T32-HL007901, K01-HL146992.

772 Examining sleep difficulties and suicide ideation among those reporting abuse and dependence on illicit drugs and alcohol

Introduction

Sleep disturbance is associated with poor mental health and may contribute to initiating or continuing use/abuse of alcohol and drugs. Using data from a nationwide survey, we examined the relationship between sleep disturbance and suicide behaviors among youth and adults, including those who report drug/alcohol use and abuse.

Methods

We analyzed data from the 2018 National Survey on Drug Use and Health (NSDUH), an annual survey collecting information about the use of illicit drugs and alcohol among non-institutionalized U.S. youth (age 12–17) and adults (age>17). The 2018 survey included 9,398 youth and 43,026 adult respondents. Depression was assessed in adults with the Kessler-6 and in youth with several questions assessing psychological distress. Those who scored at risk for psychological distress were also asked about sleep disturbance and suicidal behaviors (i.e., ideation, planning, attempt). All were asked to report their drug/alcohol use and/or abuse. Our study population included those who reported psychological distress. We conducted binary logistic regression to examine the relationship between suicidal behavior and sleep disturbance in this population. We also conducted sub-analyses to explore the relationship between suicidal behavior and sleep disturbance among those reporting drug/alcohol use and abuse.

Results

Youth were 29% male and 71% female, adults were 36% male and 64% female. Adult participants, 39% were 18 to 25, 22% were 26 to 34, and 39% were age 35 and older. Among those with psychological distress, suicidal behavior was more likely among those who reported sleep disturbance (youth: OR=2.7, 95%CI:1.8–4.0; adults: OR=1.3, 95%CI:1.2–1.5). Also, among those with psychological distress, suicidal behavior was more likely among those who reported concomitant sleep disturbance and either alcohol abuse/alcoholism (youth: OR:3.3, 95%CI:1.6–7.0; adults: OR=1.4, 95%CI:1.1–1.7); illicit drug abuse (youth: OR=3.5, 95%CI:1.6–7.4; adults: OR=1.3, 95%CI:1.0–1.6); or alcohol and illicit drug abuse (youth: OR=3.2, 95%CI:1.5–6.9; adults: OR=1.4, 95%CI:1.1–1.7).

Conclusion

Youth and adults with psychological distress and sleep disturbance are more likely to also report suicidal behaviors. Alcohol and drug use or abuse increase their risk for suicidal behavior compared to those who do not report sleep disturbance. Future work should include examination of causality and of interventions.

Support (if any)

NIH K24-HL105664, P01-AG009975, T32-HL007901, K01HL150339, 1R56HL151637

Strategy Development Pilot Study of Sleep-Restricted Operators Using Small Satellites with Displays

INTRODUCTION: Sleep restriction may lead to decreased performance and increased accidents and errors. SPHERES, a small satellite testbed, was used to examine the effects of sleep restriction and situation awareness (SA) aids on a simulation of satellite operations.METHODS: Subjects (N = 8) were trained on SPHERES, then, in a randomized order cross-over design, had 3 d of sufficient sleep (SS) or 3 d of sleep restriction (SR) before a testing session. Subjects controlled two SPHERES satellites in a space debris avoidance scenario. Dependent measures included survival time, area covered by the satellites, and satellite motion perception.RESULTS: There were significant interaction effects of sleep protocol Order (SS or SR first) and sleep Condition (SS or SR) on survival time and area covered. Post hoc tests showed longer survival time for the second testing session if the Order was SS first (Mean = 56.1 s, Median = 44.0 s) as compared to SR first (Mean = 42.7 s, Median = 33.5 s). SS-first subjects received benefit from added SA cues of the augmented display in perceiving the satellite motion.DISCUSSION: These data support that learning in a well-rested state may support development of appropriate strategies for better performance. Subjects that were SS during the first session were better able to use added SA cues provided by the augmentation and may have then developed a better mental model of the task and the system. This pilot study suggests that training guidelines for operating multiple robotic assets should permit appropriate rest before and after training to assist in mental model development and task performance.Schneider J, Saenz-Otero A, Klerman E, Stirling L. Strategy development pilot study of sleep-restricted operators using small satellites with displays. Aerosp Med Hum Perform. 2018; 89(12):1036-1044.

Identifying Objective Physiological Markers and Modifiable Behaviors for Self-Reported Stress and Mental Health Status Using Wearable Sensors and Mobile Phones: Observational Study

Background

Wearable and mobile devices that capture multimodal data have the potential to identify risk factors for high stress and poor mental health and to provide information to improve health and well-being.

Objective

We developed new tools that provide objective physiological and behavioral measures using wearable sensors and mobile phones, together with methods that improve their data integrity. The aim of this study was to examine, using machine learning, how accurately these measures could identify conditions of self-reported high stress and poor mental health and which of the underlying modalities and measures were most accurate in identifying those conditions.

Methods

We designed and conducted the 1-month SNAPSHOT study that investigated how daily behaviors and social networks influence self-reported stress, mood, and other health or well-being-related factors. We collected over 145,000 hours of data from 201 college students (age: 18-25 years, male:female=1.8:1) at one university, all recruited within self-identified social groups. Each student filled out standardized pre- and postquestionnaires on stress and mental health; during the month, each student completed twice-daily electronic diaries (e-diaries), wore two wrist-based sensors that recorded continuous physical activity and autonomic physiology, and installed an app on their mobile phone that recorded phone usage and geolocation patterns. We developed tools to make data collection more efficient, including data-check systems for sensor and mobile phone data and an e-diary administrative module for study investigators to locate possible errors in the e-diaries and communicate with participants to correct their entries promptly, which reduced the time taken to clean e-diary data by 69%. We constructed features and applied machine learning to the multimodal data to identify factors associated with self-reported poststudy stress and mental health, including behaviors that can be possibly modified by the individual to improve these measures.

Results

We identified the physiological sensor, phone, mobility, and modifiable behavior features that were best predictors for stress and mental health classification. In general, wearable sensor features showed better classification performance than mobile phone or modifiable behavior features. Wearable sensor features, including skin conductance and temperature, reached 78.3% (148/189) accuracy for classifying students into high or low stress groups and 87% (41/47) accuracy for classifying high or low mental health groups. Modifiable behavior features, including number of naps, studying duration, calls, mobility patterns, and phone-screen-on time, reached 73.5% (139/189) accuracy for stress classification and 79% (37/47) accuracy for mental health classification.

Conclusions

New semiautomated tools improved the efficiency of long-term ambulatory data collection from wearable and mobile devices. Applying machine learning to the resulting data revealed a set of both objective features and modifiable behavioral features that could classify self-reported high or low stress and mental health groups in a college student population better than previous studies and showed new insights into digital phenotyping.