Intermittent hypoxic training (IHT)

What is behind it

Oxygen deficiency stabilizes the hypoxia-inducible factor HIF-1α, which switches on a gene cascade — including erythropoietin, VEGF and enzymes of energy metabolism. This is the basis for the idea of using short hypoxic phases as a training stimulus. In the newer protocols, the hypoxia is followed not by normal air but by a hyperoxia phase with 30 to 40 % oxygen, because the alternation itself is supposed to make up the stimulus.

What of this reaches humans does not lie on the blood formation side. The only meta-analysis on clinical endpoints states verbatim that there is “no supporting evidence that IHNT/IHHT can significantly improve hematological parameters or lipid profile”. What was measured there were heart rate and blood pressure.

The same hypoxia, two different doses

The crucial point is that the same stimulus is either a training stimulus or a disease model, depending on depth and pattern. Intermittent hypoxia in the sleep apnea pattern raised mean arterial pressure in animal models across 125 papers by 13.90 mmHg (95 % CI 11.88–15.92), thickened the intima-media (absolute +5.23 µm) and enlarged atherosclerotic plaques in ApoE knockout mice.

In a second rodent meta-analysis of 92 papers, the effect on infarct size was explicitly dose-dependent in both directions: protective with mild, harmful with severe hypoxia. These are animal data with protocols that mostly model a serious disease, and not proof that a mild protocol does harm. But they are the reason why “hypoxia is a hormetic stimulus” does not hold regardless of dose.

What has come out in humans

Positive: for musculoskeletal complaints, a meta-analysis of seven studies with 321 participants found improvements in pain (standardized mean difference −0.5; p < 0.0001) and general health. The best-controlled evidence lies in a different field: in incomplete spinal cord injury, there are nine randomized, sham-controlled trials with 114 adults, with effects on walking speed, endurance and muscle strength.

Negative: the cleanest single study on whether the additional stimulus adds anything beyond normal training came out negative. 34 geriatric patients completed five to seven weeks of multimodal training, one half with hypoxia-hyperoxia, the other with a room-air sham treatment. No difference was detectable on any of the four endpoints.

What the studies show

Circulatory effects in cardiovascular patients (Glazachev et al. 2021)

14 studies, 320 patients in the intervention groups and 111 in the control groups. Positive: resting heart rate −5.35 beats per minute (p = 0.006), systolic blood pressure −13.72 mmHg. Negative: no improvement in hematological parameters, none in the lipid profile, and exercise tolerance rose within the intervention group, but not in comparison with the controls.

Added benefit on top of training, tested double-blind (Bayer et al. 2019)

34 geriatric patients aged between 64 and 92, all on a multimodal training program over five to seven weeks. The hypoxia group additionally breathed 10 to 14 % oxygen for 4 to 7 minutes, followed by 2 to 4 minutes at 30 to 40 %; the control group received a sham treatment with room air. No significant difference on any endpoint: Tinetti mobility test +14.9 % versus +15.4 % (p = 0.25), Timed Up and Go −21 % versus −26.3 % (p = 0.51), EQ-VAS +37.9 % versus +33.9 % (p = 0.24).

Tolerability of two protocols (Marzola et al. 2026)

Twelve healthy volunteers, four weeks, two sessions per week of 45 minutes each with light exercise. A training variant (30 min hypoxia, 7.5 min normoxia, 7.5 min hyperoxia) was compared with a conditioning variant (15 min hypoxia, 22.5 min normoxia, 7.5 min hyperoxia). Both were well tolerated, but the first desaturated more strongly (p = 0.048) and showed a greater rise in interleukin-6 (p = 0.021).

What the studies do not show

The advertised effect on EPO, mitochondrial density and stress resilience has no counterpart in the human literature. The meta-analysis explicitly states that there is no evidence of an improvement in hematological parameters or the lipid profile. Nothing was measured on mitochondrial density in humans. And exercise tolerance rose within the groups, not between them — that is not proof of efficacy.

One of the most cited human studies is no longer available: the pilot study in mild cognitive impairment with 21 participants was retracted in 2024. The reason was duplicated Western blot data in Figure 2, for which the authors could not provide raw material; the editors had “lost confidence in the reliability of these findings”, and the authors disagreed. In addition, the protocols are not comparable: oxygen content, number of cycles and cycle duration vary from study to study.

How to do it

What was done in the studies differs in two respects from what is often recommended. First, the oxygen content is higher: 10 to 14 % for 4 to 7 minutes in the geriatric study, six cycles of 5 minutes at 10 to 14 % and 3 minutes at 30 to 40 % in the pilot study with older adults, four cycles of 5 minutes at 12 % and 3 minutes at 33 % in cognitive impairment. Values around 9 % appear nowhere.

Second, in the intermediate phase the air breathed is not normal air but hyperoxic — that is the difference between IHT and IHHT. All studies with positive effects were conducted under medical supervision with pulse oximetry.

Safety

For this procedure, this is the most important section. In humans, the data look unremarkable: in the meta-analysis on spinal cord injury, dropout rates were low; in the one on musculoskeletal complaints, no serious adverse events were reported — but the authors note there that future studies would first have to record safety endpoints systematically. In a pilot study in nine heart failure patients, no adverse events occurred with 95 % adherence; nine people are not a safety study. At the same time, the animal data show that the dose is not a detail: on infarct size, mild hypoxia had a protective effect and severe hypoxia a harmful one. Hypoxia protocols therefore do not belong in self-directed use by people with coronary heart disease, heart failure, pulmonary hypertension, COPD, untreated sleep apnea, epilepsy, sickle cell disease or during pregnancy. No diagnosis, no treatment recommendation.

Sources

• Glazachev et al., Current Cardiology Reviews 2021
• Bayer et al., BMC Geriatrics 2019
• Shabbir et al., Journal of Back and Musculoskeletal Rehabilitation 2026
• Sogbossi et al., Clinical Rehabilitation 2026
• Marzola et al., Journal of Functional Morphology and Kinesiology 2026
• Harki et al., European Respiratory Journal 2022
• Belaidi et al., European Respiratory Review 2022
• Retraction notice, International Journal of Molecular Sciences 2024