The Caffeinated Brain Part 2: The Effect of Caffeine on Sleep-Related Electroencephalography (EEG)-A Systematic and Mechanistic Review.
Level 1 - systematic review of randomized trials
Systematic review of 32 human studies assessed with RoB 2 and ROBINS-I tools
PubMed 42075032 · doi:10.3390/nu18081220
What was done
A systematic review searched seven databases from January 1980 to January 2026 for human studies investigating caffeine administration or exposure alongside sleep-related EEG outcomes (polysomnography, spectral EEG, or derived metrics). Dual-reviewer screening and extraction were conducted, and risk of bias was assessed via RoB 2 and ROBINS-I. Due to high methodological heterogeneity across diverse dosing protocols (acute evening, daytime repeated, sleep-deprivation recovery, withdrawal, ambulatory), findings from 32 included studies were synthesized narratively.
What was found
The abstract reports qualitative directions across 32 studies without providing numerical effect sizes or confidence intervals: - Caffeine consistently suppressed low-frequency NREM EEG activity (slow-wave activity and lowest delta frequencies), particularly in early-night NREM sleep. - It attenuated the expected homeostatic rebound in low-frequency power during recovery sleep after sleep deprivation. - It frequently increased higher-frequency activity (sigma/spindle and beta ranges), reflecting a lighter, more aroused sleep profile. - REM sleep effects were less consistent, though some studies noted delayed REM timing. - Quantitative EEG measures were more sensitive than conventional sleep-stage variables at detecting disruption. - Dose, timing, habitual use, withdrawal state, age, circadian context, and ADORA2A genetic variation moderated these effects.
Why it matters
This review demonstrates that quantitative spectral EEG reveals caffeine-induced disruption to sleep depth and homeostatic recovery that standard sleep staging often misses. It directly connects adenosine receptor antagonism to impaired neurophysiological sleep restoration.
Limits
No quantitative meta-analysis was performed due to substantial methodological heterogeneity. The abstract does not report exact numerical effect sizes, total participant numbers across studies, or specific dosing thresholds, and designs varied widely across included trials.