Performance of unmodified mechanical ventilators with 2% hydrogen gas mixtures
Mancebo JG, Sack K, Nguyen R, Peng Y, Sosa S, Anders M, Roddy DJ, Kheir JN. · Respiratory Care. 2025;70(4):377–383.
Study at a glance
Preclinical
Servo-i, Servo-u, Evita Infinity V500 and Babylog VN500 ventilators connected to a closed test-lung circuit; no living subjects.
Certified premixed 2% H₂/98% O₂ and 2% H₂/21% O₂/balance N₂ were attached to ventilator O₂ and air inlets. · Ventilators were tested across 25–500 mL tidal volumes and set FIO₂ 0.21–1.0; delivered H₂ measured 1.9–2.1%.
Repeated bench runs and pre-use checks; no treatment duration or patient exposure.
Servo-i, Servo-u and Evita V500 stayed within the study's ±10% accuracy criterion for tidal volume/FIO₂ and passed checks. Babylog VN500 overdelivered volume by roughly two- to three-fold (average bias −89.2%) and consistently failed its neonatal flow-sensor check with H₂ mixtures, despite acceptable FIO₂ readings.
Bench volume-control testing only, specific ventilator models/software, no pressure-control/non-invasive/anesthesia systems and no animal/patient safety or efficacy. The authors state H₂ mixtures were not FDA-approved for human use at publication. NIH funded the work; authors disclosed no conflicts.
What kind of evidence is this?
Preclinical
Bench test of four clinical mechanical-ventilator models using a simulated patient/test lung
Critical care and ischemia-reperfusion
Other forms
Information not yet classified
Some editorial classification fields are still pending. The source-reported outcomes and result are shown below; Hydrogenology does not infer a positive or negative signal from prose automatically.
Methods
Servo-i, Servo-u, Evita Infinity V500 and Babylog VN500 ventilators connected to a closed test-lung circuit; no living subjects.
Five replicates per tidal-volume/FIO₂ combination; pre-use checks repeated three times, with the Babylog failure replicated in three separate units.
Repeated bench runs and pre-use checks; no treatment duration or patient exposure.
Certified premixed 2% H₂/98% O₂ and 2% H₂/21% O₂/balance N₂ were attached to ventilator O₂ and air inlets.
Low-concentration premixed H₂/O₂/N₂ gases — not Brown's gas; both contained 2% H₂ rather than a 2:1 H₂:O₂ electrolytic mixture.
Ventilators were tested across 25–500 mL tidal volumes and set FIO₂ 0.21–1.0; delivered H₂ measured 1.9–2.1%.
The same ventilators supplied by standard hospital wall air/O₂ without H₂.
Outcomes and reported result
Set-versus-delivered tidal volume and FIO₂, inspired H₂ concentration and pre-use-check errors.
Servo-i, Servo-u and Evita V500 stayed within the study's ±10% accuracy criterion for tidal volume/FIO₂ and passed checks. Babylog VN500 overdelivered volume by roughly two- to three-fold (average bias −89.2%) and consistently failed its neonatal flow-sensor check with H₂ mixtures, despite acceptable FIO₂ readings.
The complete free PMC article, settings, table, supplementary-delivery result and disclosures were checked for both gas mixtures, variable flow, all four ventilators, positive and device-failure findings, regulatory cautions, funding and conflicts.
A reported association, difference or mechanism is not automatically a clinical benefit.
Limitations and applicability
Bench volume-control testing only, specific ventilator models/software, no pressure-control/non-invasive/anesthesia systems and no animal/patient safety or efficacy. The authors state H₂ mixtures were not FDA-approved for human use at publication. NIH funded the work; authors disclosed no conflicts.
Device-compatibility evidence for trial planning; it does not authorize H₂ ventilation or prove clinical benefit.
No single-study GRADE certainty rating is assigned. Read how records and evidence assessments are prepared.
Sources and record status
PMID: 39969918 · DOI: 10.4187/respcare.12371
Free full article in PubMed Central.
Complete PMC article and PubMed metadata; full-text extraction checked 9 August 2026.
10 August 2026
Help people check the facts before they buy.
Hydrogenology keeps 665 source-checked records accessible without advertising or product promotion.
Help us correct the recordReport a discrepancy
Send the exact difference between this page and the linked source. The study reference is attached automatically.