Skip to the main content
Open WellbeingWhat the studies actually measured

Sleep

Caffeineand Total sleep time

Two sources here, and they say different things: one measured lower total sleep time and one found no clear difference.

How sure anyone can be

Moderate certainty, level 3 of 4

Further research may change this.

The most certain of the two findings here; each record carries its own.

The numbers reported

One source’s range, drawn against its own line of no difference and scaled to its own numbers.

-69.93-32.47 minutes4.99

This range crosses the no-difference line, so these numbers fit a real difference and no difference at all equally well.

Bar the confidence interval — the range the true effect is likely to sit inNotch the number the source reportedLine where there would be no difference at all

Two sources, newest first

We do not pick between these or average them into one number.

What we read is the published abstract, not the full paper, so a blank below may still be printed in the paper itself. A fixed rule worked this out from the source’s own fields. Nobody checked this rating by hand.

  • Systematic review, 2025

    Moderate certainty, level 3 of 4

    The Effect of Consuming Caffeine Before Late Afternoon/Evening Training or Competition on Sleep: A Systematic Review with Meta-Analysis.

    Sports (Basel). 2025;13(9). doi:10.3390/sports13090317. PMID:41003623.

    Number reported
    -32.47 minutes
    Which way it went
    No clear difference
    Likely range
    -69.93 to 4.99
    Source’s own words
    minutes of total sleep time, mean difference vs control
    Who was studied
    Athlete
    Compared with
    control arm of the included trials
    Amount taken
    3 to 6 mg/kg body mass, taken before late-afternoon or evening training
    For how long
    Not in the abstract
    Disagreement between studies
    I2 = 0% for the total-sleep-time pooling
    Kind of record
    Journal article
    How this rating was reachedstarts at High, one step down: imprecision
    1. Starts at High
    2. Imprecision brought it down to Moderatethe confidence interval includes no effectReported in the abstract
    3. Ends at ModerateOne step down.
    Checked, nothing to step it down for
    risk of bias and indirectness
    Inconsistency
    No disagreement reported, and no count of studies behind it

    Imprecision stops at the first check that applies; others may apply too.

    AI-reviewed & signed — an AI agent re-read this finding against the saved source and checked that the quote, the numbers, the units and the direction match it. No person read it. The owner authorised the agent to sign the batch with the owner's key, and every build rechecks that signature against these exact bytes.

    Read the abstract on PubMed

  • Systematic review, 2025

    Low certainty, level 2 of 4

    Age- and dose-specific effects of caffeine on sleep: A meta-analysis of controlled crossover trials.

    Sleep Med. 2025;136:106874. doi:10.1016/j.sleep.2025.106874. PMID:41124973.

    Number reported
    -34.67 minutes
    Which way it went
    Lower
    Likely range
    Not in the abstract
    Source’s own words
    minutes of total sleep time, mean difference vs placebo
    Who was studied
    Adult
    Compared with
    placebo
    Amount taken
    Not in the abstract
    For how long
    Not in the abstract
    Disagreement between studies
    Not in the abstract
    Kind of record
    Journal article
    How this rating was reachedstarts at High, 2 steps down: risk of bias and imprecision
    1. Starts at High
    2. Risk of bias brought it down to Moderatethe abstract reports no risk-of-bias or quality appraisal, so study limitations cannot be ruled outNot in the abstract
    3. Imprecision brought it down to Lowthe abstract reports no confidence interval for the estimateNot in the abstract
    4. Ends at Low2 steps down.
    Checked, nothing to step it down for
    indirectness
    Inconsistency
    No disagreement reported, and no count of studies behind it

    Imprecision stops at the first check that applies; others may apply too.

    AI-reviewed & signed — an AI agent re-read this finding against the saved source and checked that the quote, the numbers, the units and the direction match it. No person read it. The owner authorised the agent to sign the batch with the owner's key, and every build rechecks that signature against these exact bytes.

    Read the abstract on PubMed

What is still missing from these readings

  • One gives no likely range, so how precise that number is is unknown.
  • Two do not say how many people took part.
  • Amount taken — One of Two readings: Not in the abstract.
  • For how long — Not in the abstract.

Something nobody wrote down is never a zero and never a guess.

Still to read for Caffeine

Ten of the 19 things on our list for Caffeine have no reading here yet:

  • Anxiety
  • Cognitive performance
  • Estimated VO2max
  • Fatigue
  • Heart rate variability (RMSSD)
  • Light sleep
  • Number of awakenings
  • Resting heart rate
  • Serum cortisol (morning)
  • Wake after sleep onset (WASO)

Still to read means nobody here has got to it, not that it was tried and did nothing or that no research exists. A pairing on no list is not a gap: we never set out to read it.

Read next

This is a library, not advice. No recommendation, no dose, no diagnosis, no ranking. We never ask about your symptoms, and nothing here is tailored to you.