Safety and interactions: the data record and its limits

Not legally assertable: This article makes no assurance about safety, tolerance or harmlessness, neither in general nor with a qualification, neither in its own words nor as a rendering of an authority or a study summary. It gives no usage advice, names no amount and no timing. Nor does it release or exclude any group of people. What follows is a description of recording procedures and their limits, and nothing beyond that. The wider frame is given by the overview of evidence levels and research limits.

Evidence status M3: Studies record events under a pre-specified protocol: it fixes who is asked or examined over which period, which events are documented at all and how they are classified. An event recorded that way is an observation inside that frame. The CONSORT statement: Schulz, Altman and Moher (2010), BMJ sets out which of these details a randomised study has to report. Limit: it is a reporting standard and carries no statement about molecular hydrogen itself.

Evidence status M3: An event that occurs rarely or becomes visible only after a long time cannot systematically be found in a small and short study. That is not a property of the substance under study but a property of the study design. From this follows the central point of this article: that something was not found in a study is no proof that it does not exist. A non-finding and proof of absence are two different things.

Evidence status H1: Gaboreau et al. (2024), Journal of Clinical Medicine report a randomised, triple-blinded, placebo-controlled phase 3 study in outpatients with COVID-19 and document the adverse events recorded in it. Limit: that is a disease population over a defined period; the work describes its own recording and establishes no statement about safety for healthy people. A second example from a clinical context is Sumbalová et al. (2023), International Journal of Molecular Sciences. Limit: a clinical setting with laboratory parameters in which molecular hydrogen was studied alongside ongoing clinical care; referred to here solely as a study description and not as a benefit. How clinical populations are bounded in another field is shown in clinical populations in metabolic research.

Evidence status N: For interactions with medicines, no suitable primary source is documented in this article. That gap is named and is not filled instead with reviews, personal reports or supplier information. The question therefore stays open: no interaction is described, and none is ruled out. How studies are read by population and design is set out in reading studies by population and design.

Evidence status N: Individual questions about medicines, about a pregnancy or about existing conditions are answered in a professional consultation that knows the person concerned and their records. This article names that responsibility and attaches to it neither a recommendation nor a warning nor the exclusion of any group of people. Why no drinking amount appears in this context either is set out in why no drinking amount appears here. How the acid-and-base question is placed as a measurement is shown in acid-and-base questions as measurements.