Living in biological darkness III: Effects of low-level pre-midday lighting on markers of depression in healthy subjects.
Level 2 - randomized trial
Randomized controlled trial
PubMed 41240877 · doi:10.1016/j.jpsychires.2025.11.008
What was done
Twenty healthy adults (mean age 24.5 ± 2.8 years; 10 females) were randomized into two parallel groups of 10 and exposed to either low-intensity incandescent lighting (2700K, 55 lx, 12 mlux) or higher-intensity fluorescent lighting (3500K, 800 lx, 481 mlux) across five mornings (8:00–12:00) over a 7-day period. Assessments conducted before and after the intervention included two-night polysomnography, 32-hour fractionated urinary cortisol, salivary cortisol, salivary melatonin, reaction time, subjective daytime sleepiness, and subjective sadness.
What was found
Low-level pre-midday lighting significantly resulted in: - Increased evening urinary cortisol from 19:00 to 23:00 (p = .021) - Increased afternoon salivary cortisol at 16:00 (p = .026) - Reduced total sleep time by approximately 25 minutes (p = .016) - Shifted slow-wave activity from the second to the third NREM-REM cycle (p = .018) - Increased subjective sleepiness (p = .023) and sadness (p = .016) Higher-intensity lighting resulted in increased REM sleep toward the end of the sleep period. Circadian phase remained unchanged across both groups. Baseline and endpoint absolute quantitative values for most markers were not reported in the abstract.
Why it matters
This study shows that insufficient daytime light exposure can acutely disrupt HPA-axis regulation, alter sleep architecture, and increase sadness in healthy individuals, mimicking physiological features of depressive disorders.
Limits
The sample size was very small (n = 10 per group), creating substantial risk of imprecision. The intervention was brief (5 mornings), leaving long-term durability and adaptation unknown. The study was conducted strictly in young, healthy volunteers, limiting generalizability to clinical populations. Specific baseline and endpoint values were largely omitted from the abstract.