Pattern 06 / 12 Otto observes
Coffee after 2pm → fragmented sleep at night
You think a 4pm coffee does not affect you. Polysomnography says otherwise. You tend to lose about an hour of sleep — without noticing.
A dose of 400 mg of caffeine (≈ 3 espressos) taken 6 hours before bedtime reduces total sleep duration by over an hour — measured objectively by polysomnography. Subjectively, many participants report no difference. The closer the coffee is to bedtime, the more sleep fragmentation rises.
Drake et al. (2013) · J Clin Sleep Medicine
A 400 mg caffeine dose taken at 0, 3 and 6 hours before bedtime significantly reduced sleep duration and efficiency vs. placebo. Even 6 hours before bedtime, total sleep dropped by more than an hour — without participants consciously reporting the difference.
Gardiner et al. (2025) · Sleep
Crossover RCT (n=23, dose-timing): 400 mg taken within 12 hours of bedtime altered sleep architecture; under 8 hours, fragmentation rose. The low dose (100 mg) had no significant effect even at 4 hours before bedtime.
Gardiner et al. (2023) · Sleep Medicine Reviews
Meta-analysis of 24 controlled trials: caffeine reduced total sleep by an average of about 45 minutes and sleep efficiency by about 7%, with less slow-wave sleep and more light sleep. To avoid sleep loss, the authors suggest a cut-off of at least 8.8 hours between caffeine (107 mg) and bedtime.
Caffeine blocks adenosine receptors — the molecule that accumulates in the brain over the day and produces the feeling of sleepiness. When the block holds into the evening, sleep pressure is there, but the signal is cut.
You still fall asleep — because the tiredness is real — but the sleep is shallower, more fragmented, with less slow-wave (deep) sleep. You wake up feeling like you slept, but without the feeling of having recovered.
Caffeine half-life varies between 2 and 10 hours across adults. If you are a slow metaboliser, a 2pm coffee is still active at 10pm.
The time and dose of coffee entries (NutriBase) + sleep architecture from your wearable: total time, efficiency, number of awakenings, NREM/REM ratio (Apple Health / Health Connect).
You do not have to cut out coffee. Move the cut-off to 2pm and watch in Otto what changes in your sleep over the next 7–10 days. If you are a slow metaboliser (you will see fragmentation persist), try noon.
The dose matters as much as the timing. Under 100 mg (≈ a single espresso) is a much safer zone than 400 mg in the evening.
Otto observes patterns like this in your data. Launching soon.
Notify me →