REF-1675 · Jiang et al., 2026 · Scientific Reports, · Paper · PubMed · Free full text

Training intervals in simulated altitude beat the same training at sea level by a moderate standardised amount for aerobic capacity in athletes: in a network meta-analysis of 29 randomised trials (603 trained athletes, blocks of about 4 weeks at a simulated 2,100-5,000 m), interval hypoxic training gave SMD 0.53 (95% CI 0.18-0.88) and repeated sprints in hypoxia 0.40 (0.10-0.70) for aerobic capacity, sprint intervals in hypoxia 0.26 (not significant); only repeated sprints in hypoxia raised anaerobic capacity (0.55, 0.29-0.81). No method was shown to beat another head to head (very low to moderate certainty). Read in full

Main caveat: The authors': athletes from many sports and levels, and different intervention lengths and tests, may threaten transitivity.
fairly strong evidenceMeasured on trained to highly trained athletesRead in full

Jiang and colleagues asked which of three ways of training in hypoxia works best for athletes, against the same training done in normal air at matched intensity: interval hypoxic training (efforts of about 30 s to 5 min with similar recoveries), repeated-sprint training in hypoxia (sprints of 10 s or less with recoveries of 60 s or less) and sprint interval training in hypoxia (20-30-s all-out efforts with 3-5 min recovery).

They searched four databases to November 2025 and included 29 randomised trials published 2005-2025 with 603 trained athletes from team, endurance and individual sports: 10 IHT, 14 RSH and 5 SIH trials, simulated altitudes of 2,100 to 5,000 m (most often 2,500-3,500 m), blocks averaging 4.3 weeks with 2-4 sessions a week.

Against normoxic training, IHT and RSH improved aerobic capacity in both the pairwise (SMD 0.55 and 0.43) and the network analysis (0.53 and 0.40), while SIH did not; for anaerobic capacity only RSH improved (pairwise 0.57, network 0.55).

The probability ranking put IHT first for aerobic and RSH first for anaerobic outcomes, but no direct comparison between the hypoxic methods was significant, and GRADE rated those comparisons very low to moderate. Exploratory altitude subgroups found aerobic gains significant at moderate and high but not low simulated altitude, and anaerobic gains only at moderate altitude.

The authors present the ranking as direction for prescription that needs confirming in head-to-head trials. Our reading: the finding that holds in both analyses is that interval or repeated-sprint work in hypoxia added a moderate standardised amount to aerobic capacity over the same work at sea level; the ranking between methods is not established.

Participants

29 randomised trials, 603 trained athletes of both sexes (the sex split is not given in the text); team sports (football, rugby, basketball), endurance sports (running, swimming, cycling) and other individual sports.
Ages, training volumes and competition levels by trial are in Appendix S3, which we did not read.

Protocol

Search: PubMed, Web of Science, Embase and Cochrane Library from inception to 1 November 2025, plus reference lists of reviews; 4,937 records, 3,693 after removing 1,244 duplicates, 96 full texts, 29 included.
Inclusion: randomised controlled trials in trained athletes; IHT, RSH or SIH against intensity-matched normoxic training; aerobic or anaerobic capacity as outcomes. Excluded: acute (single-session) studies, other hypoxic methods, no normoxic control.
Interventions: 10 IHT, 14 RSH, 5 SIH trials; simulated altitude 2,100-5,000 m (most often 2,500-3,500 m); mean duration 4.3 weeks (SD 2.1), 2-4 sessions a week.
Outcomes: aerobic capacity mostly from incremental treadmill or cycle tests (VO2max or peak power); anaerobic capacity from Wingate tests, repeated-sprint-ability protocols and sport-specific time trials.
Analysis: pre-to-post change scores, change SDs imputed with r = 0.5 (sensitivity at r = 0.3 and 0.7); pairwise random-effects SMDs in Review Manager 5.3; frequentist random-effects NMA in Stata 16 with a common heterogeneity parameter; consistency by loop-specific tests, a global inconsistency model and node-splitting; SUCRA ranking; Begg's and Egger's tests and comparison-adjusted funnel plots.Altitude subgroups: low (>1,500-2,500 m), moderate (>2,500-3,500 m), high (>3,500 m), pairwise only.
Risk of bias: original seven-domain Cochrane tool by two reviewers; participant blinding excluded from the overall score; overall low (no high-risk domain and three or fewer unclear), moderate, or high.
Funding and interests: the authors declare no competing interests; the text carries no funding statement.

Key Results

Hypoxic training against intensity-matched normoxic training (SMD, 95% CI):

OutcomeMethodPairwiseNetwork
Aerobic capacity (28 trials, 537 athletes pairwise)IHT0.55 (0.25-0.85), I² 0%0.53 (0.18-0.88)
RSH0.43 (0.01-0.84), p = 0.043, I² 60.1%0.40 (0.10-0.70)
SIH0.25 (−0.15 to 0.66), not significant0.26 (−0.18 to 0.71), not significant
Anaerobic capacity (24 trials, 474 athletes pairwise)RSH0.57 (0.30-0.84), I² 23.5%0.55 (0.29-0.81)
IHT0.13 (−0.77 to 1.03), not significant, I² 67.2%0.14 (−0.42 to 0.70), not significant
SIH0.04 (−0.42 to 0.50), not significant0.11 (−0.31 to 0.53), not significant
Head to head (network): aerobic, IHT against RSH 0.13 (−0.33 to 0.59), IHT against SIH 0.27 (−0.29 to 0.83), RSH against SIH 0.14 (−0.38 to 0.66); anaerobic, RSH against SIH 0.44 (−0.04 to 0.92), IHT against RSH −0.41 (−1.02 to 0.21). None significant.
Ranking (SUCRA): aerobic IHT 84.2, RSH 66.5, SIH 45.0, control 4.4; anaerobic RSH 95.6, IHT 44.0, SIH 39.9, control 20.5. The authors note SUCRA conveys neither size nor certainty.
Simulated altitude (pairwise, exploratory): aerobic, low 0.44 (−0.01 to 0.88, not significant), moderate 0.43 (0.13-0.73), high 0.47 (0.02-0.92); anaerobic, low 0.52 (−0.76 to 1.81), moderate 0.53 (0.30-0.77, I² 20.1%), high −0.38 (−0.89 to 0.13); between-subgroup difference for anaerobic p = 0.006.
Risk of bias: overall low in 17 trials (58.6%), moderate in 12 (41.4%), high in none; allocation concealment unclear in all 29; random sequence low risk in 24, unclear in 5; outcome-assessor blinding low in 19, unclear in 10; "other bias" high in 12 (41.4%).
Certainty (GRADE): high for RSH against control (both outcomes); moderate for IHT and SIH against control (downgraded for inconsistency and publication bias respectively); very low for IHT against RSH and IHT against SIH; moderate for RSH against SIH and for the overall ranking.
Sensitivity: leave-one-out and r = 0.3 or 0.7 left direction and significance unchanged; no inconsistency in either network (aerobic loop p = 0.816, global p = 0.79; anaerobic loop p = 0.78, global p = 0.79); no publication bias detected by funnel plots, Begg's and Egger's tests.
Internal inconsistencies (our reading): the discussion calls the aerobic gain significant at low, moderate and high altitude, but the low-altitude interval crosses zero; GRADE rates RSH against control high although its pairwise aerobic I² is 60.1%; the limitations call the head-to-head evidence "mostly low certainty" where the GRADE section says very low to moderate; IHT is described as "high-intensity (70% VO2max)", an intensity that is not high on most scales.

Practical Application

A few weeks of interval or repeated-sprint sessions in simulated altitude added a moderate standardised amount to aerobic capacity over the same sessions at sea level (network SMD 0.53 for intervals, 0.40 for repeated sprints, in trained athletes, blocks of about 4 weeks with 2-4 sessions a week). The size is in standard deviations, not watts or mL/kg/min, and the outcomes mix VO2max and peak power.
For an endurance athlete, hypoxic intervals ranked first for aerobic capacity, but they were not shown to beat repeated sprints in hypoxia directly (0.13, −0.33 to 0.59; very low certainty). Sprint intervals in hypoxia (20-30 s all-out, long rests) did not improve either outcome.
Repeated short sprints in hypoxia were the only method that raised anaerobic capacity (0.55); in an exploratory pairwise subgroup across the methods, anaerobic gains were significant only at moderate simulated altitude (above 2,500 up to 3,500 m). That suits team-sport and repeated-effort demands more than steady endurance.
Where this differs from the guide: for an athlete without altitude residence SYNTH-06 says not passive hypoxia, citing a narrative review that rates hypoxic training above passive exposure (REF-209), and names heat for the blood adaptation, but it gives no prescription for training sessions in hypoxia; this analysis adds that interval or repeated-sprint sessions in simulated altitude (about 4 weeks, 2-4 sessions a week) improved aerobic capacity in trained athletes over the same sessions at sea level, an option the guide's row does not offer.It does not compare hypoxic sessions with heat, measures no haemoglobin or race outcome, and does not touch the guide's iron, live high-train low dose or few-per-cent lines.A second 2026 meta-analysis found no significant VO2max gain from training in hypoxia in athletes (REF-1677, live-low train-high), so the two do not agree on athletes; the outcomes differ (VO2max alone there, VO2max or peak power as standardised differences here) and neither measured race performance.
Inference, not tested here: the gain reported is over identical sea-level training, so it is an add-on for athletes who already train consistently and have the equipment; whether it lasts after the block or shows up in a race was not studied.

Limitations

The authors': athletes from many sports and levels, and different intervention lengths and tests, may threaten transitivity; composite outcomes under one SMD may hide method-specific effects; altitude subgroups had few trials each and were not run in the network; publication-bias tests are underpowered with few trials per comparison; the head-to-head comparisons are largely indirect.
Change-score SDs were imputed with an assumed correlation (r = 0.5), though sensitivity analyses at 0.3 and 0.7 did not change the conclusions.
Allocation concealment was unclear in every trial, and participant blinding, which is hard in hypoxic training, was left out of the overall rating; a placebo effect of knowing one trains "at altitude" is not excluded.
SIH rests on five trials and is a singly connected node in the network.
No follow-up after the blocks: durability of the gain is unknown. No race or time-trial outcome for endurance athletes is reported separately.
The text has internal inconsistencies (listed under Key Results). We did not read the appendices (study table, risk-of-bias summary, GRADE profile, forest plots).
How this entry was checked and changed
Why this level

a meta-analysis of randomised trials whose risk of bias we read in the full text (17 of 29 low, 12 moderate, none high; GRADE high for RSH and moderate for IHT and SIH against control), held at A- because the trials are mostly small (603 athletes over 29 trials), the outcomes are composites of different tests, the head-to-head comparisons rest on indirect evidence of very low to moderate certainty, and the altitude subgroups are exploratory; aerobic capacity (VO2max or peak power) is a capacity outcome, so no marker step