REF-1503 · de Carvalho e Silva et al., 2022 · Sports Medicine - Open · Paper · PubMed · Free full text

Does a leg session hurt the next run?

Across 13 pooled studies in male runners, peak torque fell and soreness, CK and RPE rose; VO2, HR, lactate: no significant difference. Performance narrated: 3 of 4 time-to-exhaustion studies fell.

Main caveat: Performance was not pooled. The question athletes ask has only a short narrative, and the abstract overstates it in two places (time to exhaustion, speed).
moderate evidenceMeasured on recreational to highly trained athletes

The review asked what a single strength session does to a later run. From 6,532 records, 19 studies met the criteria and 13 went into meta-analyses, with sessions ranging from damage protocols and drop jumps to traditional heavy lifting, combined strength-and-running sessions and potentiation warm-ups.

Peak torque, soreness, creatine kinase and RPE changed for the worse after strength work. Jump height, VO2, heart rate, lactate, ventilation and RER did not differ on pooling.

Performance was too scattered to pool: of four time-to-exhaustion studies three showed less time and one no significant difference, of two speed studies one showed a change, a potentiation protocol sped up a timed run, and one 5-km time worsened after 30 min and, less so, after 48 h.

Our reading: the muscle clearly carries the session; the heart and lungs at submaximal speed mostly do not show it on pooling. That does not mean the run is unaffected: harder and longer tests were the ones that fell, and the windows were too mixed to give a safe gap.

Protocol

Participants: n = 238 across the 19 studies (our sum of Table 2; 8 to 24 per study; the review gives no total, and some studies may share runners).
Search and selection: PubMed and Scopus, 1995 to April 2021, English; 6,532 records, 746 duplicates, 5,786 screened, 48 full texts, 19 included, 13 in meta-analyses. Two independent reviewers, a third on conflict. No registration reported.
Eligibility: a lower-limb strength session above 15 min, any load, then running performance in recreational or trained runners, tests longer than 75 s; comparison with a control group or with a time before the session; men. Training programmes (more than one session) excluded.
Included studies: 4 aimed at maximum muscle damage, 2 at potentiation, 1 plyometrics with strength, 5 combined with aerobic work, 7 traditional strength.
Time points: 10 studies immediately after, 11 also at 24 h, 9 also at 48 h; some at 6 h.
Quality: Joanna Briggs checklist of 8 questions; all 19 rated "reasonable" (62.5-75%); every study scored no on questions 5 and 6 (confounders and how they were handled).
Analysis: Review Manager 5.3; effects as standardized mean difference with 95% CI; heterogeneity by Chi-squared, tau-squared and I2.
Funding and interests: no funding; no competing interests declared.

Key Results

Pooled comparisons, strength session vs control, as printed:

Variable"SMD" [95% CI]pReading
Peak torque41.78 [14.50 to 69.05]0.003Lower after strength
DOMS−3.90 [−4.37 to −3.44]< 0.0001Higher after strength
Creatine kinase−80.18 [−110.17 to −49.39]< 0.0001Higher after strength
RPE−0.56 [−0.79 to −0.33]< 0.0001Higher after strength
Countermovement jump0.48 [−1.5 to 2.45]0.64No difference
Lactate−0.31 [−0.71 to 0.09]0.13No difference
Heart rate−2.84 [−6.07 to 0.40]0.09No difference
VO2−0.05 [−0.30 to 0.20]0.70No difference
Ventilation−4.38 [−10.02 to 1.24]0.13No difference
RER−0.01 [−0.02 to 0.01]0.33No difference

These numbers do not add up as standardized effects. An SMD of 41.78 or −80.18 is not possible in practice; those and the heart-rate and ventilation values look like raw mean differences in the original units (our inference). Use the direction and the p, not the size. The authors write that heterogeneity and inconsistency were low; the I2 values are in the forest plots, which are images and were not read.

Performance, narrative only (not pooled; means and SDs missing, single studies, or different intervals):

Time to exhaustion: four studies; three, with traditional strength sessions, showed a reduction, one no significant difference. The abstract says "all".
Speed: two studies evaluated it; only one changed significantly (13.9 ± 1.7 to 13.6 ± 1.7 km/h). The abstract says a reduction in two.
Timed runs: a potentiation protocol shortened the timed run (Low); 5 km went from 20.63 ± 2.42 min before to 22.40 ± 2.86 min 30 min after and 21.26 ± 2.56 min 48 h after (Guimarães, p = 0.02 and p = 0.04). The Discussion calls the 48-h difference "irrelevant" although the Results give it a significant p.
Stride length: reduced in two of three studies. Distance covered: no difference in one study (11.6 ± 0.9 vs 11.7 ± 1.0 km).

Mechanisms

Measured: peak torque, soreness, creatine kinase and RPE worsened; submaximal VO2, heart rate, lactate and ventilation showed no pooled difference.
Authors' interpretation: runners' performance after strength sessions seems not to be mediated by cardiorespiratory or metabolic responses, with muscle damage, neural fatigue and altered mechanics as candidates (they cite these from earlier work, and did not test them).
Inference: pooling immediate, 6-h, 24-h and 48-h tests after damage protocols, potentiation warm-ups and ordinary lifting can cancel real effects. Table 2 lists raised VO2 in the Burt, Marcello, Drummond and Doma 2020 studies, mostly big or immediate doses. The authors could not separate time points. So "no pooled difference" does not license a hard run the next morning.

Practical Application

Does a leg session hurt the next run? Across 13 pooled studies in male runners, peak torque fell and soreness, CK and RPE rose; VO2, HR, lactate: no significant difference. Performance narrated: 3 of 4 time-to-exhaustion studies fell.
Good for: telling an athlete why an easy run after legs can feel bad without a clear rise in measured cost: the muscle markers and effort rose, the pooled heart and lung responses showed no significant difference, with time points mixed (B+ for markers).
Not good for: a time gap. The authors could not analyse immediate vs 6, 24 or 48 h. Gaps stay with SYNTH-31, which is practice read off small time-course studies.
Hard efforts look more exposed than easy ones (our reading): the falls were mostly in time to exhaustion and one 5 km, tested from 30 min to 48 h after the session; the authors recommend alternating-day strength sessions, their reading, not tested here.
HYROX (transfer): not from this paper; the base's own practice: keep the heavy lift away from the key run (SYNTH-71, SYNTH-31). A station inside a race is a different stimulus and this review does not measure it.

Limitations

Performance was not pooled. The question athletes ask has only a short narrative, and the abstract overstates it in two places (time to exhaustion, speed).
Effect sizes mislabelled. See Key Results. Direction only.
Mixed windows, protocols and comparators. Damage protocols with potentiation warm-ups and traditional sessions, control groups with before-session baselines; the authors state that the study count did not allow a robust analysis by time point.
Men only by eligibility, yet the review lists Marcora with n = 24 where the base's REF-1298 gives 30 (24 men, 6 women): it appears to have counted the men while the trial's results cover everyone.
Search and quality. Two databases, English only, no registered protocol reported, no certainty grading; all studies "reasonable", all scoring no on the confounder questions. The funnel plot is read as "low risk of bias" on few studies with these mislabelled effects.
Internal inconsistency. The Conclusion and Key Points say low impact, the Practical Implications say "a substantial deleterious effect". Reference numbers differ between Table 1, Table 2 and the text; Table 1 lists Doma 2019 and Table 2 Doma 2020.
How this entry was checked and changed
Why this level

meta-analysis by design (rubric A-), held one step down: what was pooled are markers, performance is narrated only, the effect sizes are mislabelled, recovery windows are mixed, and the review is restricted to men. Read from the full text (CC BY). The performance narrative on its own is B-level at most

Full text (checked 2026-10-30)
Of 5786 articles screened for eligibility the full texts of 48 potentially eligible studies were evaluated; 19 met the criteria and were included in the review, among which 13 made up the meta-analyses (full text)
Peak torque: SMD [95% CI] = 41.78 [14.50 to 69.05], Z = 3, p = 0.003; creatine kinase: SMD [95% CI] = − 80.18 [− 110.17 to − 49.39] (full text)
VO2: SMD [95% CI] = − 0.05 [− 0.30 to 0.20], p = 0.70; lactate: − 0.31 [− 0.71 to 0.09], p = 0.13; RPE: − 0.56 [− 0.79 to − 0.33] (full text)
The time to exhaustion was analyzed in four studies; in three there was a reduction in time for the experimental group, whereas in one there were no significant differences (full text)
5 km: 20.63 ± 2.42 min before intervention; 22.40 ± 2.86 min after 30 min and 21.26 ± 2.56 min after 48 h (full text)