REF-1655 · Almeida et al., 2024 · Current Cardiology Reviews, · Paper · PubMed

Caffeine before exercise did not change how heart rate variability recovered afterwards: across 7 pooled placebo-controlled trials in active adults, post-exercise RMSSD (SMD −0.03) and HF power (SMD −0.061) matched placebo for doses of 2.1-6.0 mg/kg by capsule, drink or gum, with no dose relation and moderate certainty

Main caveat: Trial sizes, participants' sex and fitness, exercise modes and per-trial risk of bias are not in the abstract.
moderate evidenceMeasured on recreational to trained athletesRead from the abstract only

Almeida and colleagues asked whether the way caffeine is taken before exercise (its form, timing and dose) changes heart rate variability in the recovery period afterwards. They searched MEDLINE, Web of Science, LILACS and SCOPUS to May 2023 for randomised trials in active adults that compared caffeine with placebo and measured HRV before and after exercise.

Ten studies met the criteria and seven were pooled. The trials used capsules, drinks or gum, taken 10, 45 or 60 min before exercise, at 2.1 to 6.0 mg/kg. Neither vagal index differed from placebo after exercise: RMSSD (standardised mean difference −0.03) and high-frequency power (−0.061). A fixed-effects meta-regression found no link between dose and HRV.

The authors conclude, with what they call moderate-certainty evidence, that these caffeine strategies do not significantly change post-exercise HRV recovery, read as vagal modulation.

Participants

Active adults aged 18 and over (inclusion criterion). Total number, sex, training level and caffeine habits: not in the abstract.

Methods

Search: MEDLINE, Web of Science, LILACS and SCOPUS, ending May 2023; PRISMA; registered in PROSPERO (CRD42023425885).
Inclusion: randomised clinical trials comparing caffeine with placebo on post-exercise HRV in active adults; HRV measured before and after exercise as the primary endpoint.
Studies: 10 in the systematic review, 7 in the meta-analysis.
Strategies covered: capsule, drink or gum; 10, 45 or 60 min before exercise; 2.1-6.0 mg/kg.
Analysis: pooled standardised mean differences; meta-regression of HRV on dose with a fixed-effects model.
Quality: the conclusion is rated moderate certainty; the risk-of-bias tool and per-trial judgements are not in the abstract.

Key Results

Post-exercise HRV indexCaffeine vs placebo, SMD (95% CI)p
RMSSD−0.03 (−0.265 to 0.197)0.77
High-frequency (HF) power−0.061 (−0.272 to 0.150)0.57
Dose: no significant relation between the caffeine dose and HRV indices in the meta-regression (p > 0.05).
Certainty of the no-effect conclusion: moderate, by the authors' rating.

Practical Application

Caffeine before a session is not a reason to discount the HRV recovery seen after it. In 7 pooled placebo-controlled trials, 2.1-6.0 mg/kg by capsule, drink or gum, taken 10 to 60 min before exercise, left post-exercise RMSSD and HF power where placebo left them, and a larger dose did not change that.
This is about the marker, not about recovery itself. HRV is a proxy for autonomic state (PHYS-44); the review shows caffeine did not shift that proxy after exercise, not that it speeds or slows the return of performance.
Inference, not tested here: a dose late in the day can still reach the next morning's HRV through sleep (REF-973, SYNTH-07); this review looked at the post-exercise period only.
Not settled by this design: overnight or next-morning HRV, doses above 6 mg/kg, habitual heavy users against non-users, women against men, and performance in the hours after; none is reported in the abstract.

Limitations

Trial sizes, participants' sex and fitness, exercise modes and per-trial risk of bias are not in the abstract; 7 pooled trials is a small base for a meta-regression on dose.
HRV is a marker; no performance, sleep or next-day outcome was pooled.
The time after exercise at which HRV was read is not stated in the abstract.
Doses reached 6.0 mg/kg at most; higher doses were not covered.
Written from the abstract and the PubMed record; full text not read.
How this entry was checked and changed
Why this level

the abstract reports the certainty of its conclusion (moderate), which lifts the abstract-only cap (rule 1); a meta-analysis of controlled trials, but the abstract gives no trial sizes, and its pooled 95% CIs (about 0.42-0.46 SMD wide) fit roughly 145-170 participants across the 7 pooled trials, about 20-25 a trial or fewer (our estimate), so mostly small trials (A-); one step down because the outcome, HRV, is a marker (rule 3); per-trial risk of bias is not in the abstract

Content correction 2026-11-07
2026-11-07: added from the abstract and the PubMed record; level B+ (meta-analysis of placebo-controlled trials whose abstract reports moderate certainty; mostly small trials by an estimate from the pooled CIs (A-); one step down for an HRV outcome).