Tubbs · The Journal of clinical psychiatry 2024 · Cross-sectional surveillance study · n=129,173 violent deaths (78,647 suicides and 50,526 homicides)

Risk for Suicide and Homicide Peaks at Night: Findings From the National Violent Death Reporting System, 35 States, 2003-2017.

Cited 6 times in the scientific literature.

Level 4 - case-series / case-control

Cross-sectional population surveillance study linking death records with aggregate time-use survey data

PubMed 38814111 · doi:10.4088/JCP.23m15207 · record verified 2026-08-29

What was done

Researchers analyzed data from the National Violent Death Reporting System (NVDRS) across 35 US states from 2003 to 2017, examining 78,647 suicides and 50,526 homicides. To account for the proportion of people awake at different hours, mortality data were combined with population wakefulness data from the American Time Use Survey. Hourly adjusted incident risk ratios (aIRRs) were estimated, and two-way ANOVAs with post hoc analyses were used to test time-by-subgroup interactions across demographic and behavioral variables.

What was found

Unadjusted counts peaked at 12:00 PM for suicide and 10:00–11:00 PM for homicide. After adjusting for demographics and population wakefulness, suicide risk peaked at 3:00 AM (aIRR 5.20, 95% CI [4.74, 5.70]) and homicide risk peaked at 2:00 AM (aIRR 8.04, 95% CI [6.35, 10.2]). Hourly risk for suicide varied by age, ethnicity, blood alcohol level, and current partner conflict; hourly risk for homicide varied by sex and blood alcohol level. Nighttime risk was higher among young adults and those with alcohol intoxication, but not among individuals with a documented history of suicidal ideation or attempts.

Why it matters

This study provides large-scale empirical evidence for the Mind after Midnight hypothesis, demonstrating that nocturnal wakefulness substantially increases the relative risk for fatal dysregulated behaviors, including both suicide and homicide.

Limits

Wakefulness was estimated at the population level via time-use surveys rather than measured directly in individuals prior to death. The study is observational and restricted to 35 US states. nvDRS records may suffer from incomplete or inconsistent post-mortem reporting for variables such as toxicology, prior psychiatric history, and situational stressors.

Cited by