Pilot study of H₂ therapy in Parkinson's disease: a randomized double-blind placebo-controlled trial
Yoritaka A, Takanashi M, Hirayama M, Nakahara T, Ohta S, Hattori N. · Movement Disorders. 2013;28(6):836–839.
What kind of evidence is this?
Human
Randomized double-blind placebo-controlled parallel-group pilot trial
Parkinson’s disease
H₂-rich water
Not reported in this record.
Not reported in this record.
Methods at a glance
Japanese patients with Parkinson's disease receiving stable levodopa and other antiparkinsonian medication.
18 participants were randomized, nine per group. One placebo participant discontinued because of pollakiuria, leaving H₂-water n=9 and placebo n=8 at week 48.
Forty-eight weeks, with assessments at baseline and during follow-up through week 48.
Participants drank 1,000 mL/day of hydrogen-rich water in addition to their unchanged antiparkinsonian medication.
H₂ dissolved in drinking water — H₂ only, not Brown's gas.
The article describes saturated hydrogen water at approximately 0.8 mmol/L, supplied as a total daily volume of 1,000 mL.
Not applicable — drinking water, not gas inhalation.
No O₂ was co-delivered as part of the intervention.
Pseudo/placebo water, 1,000 mL/day, with the same stable levodopa and antiparkinsonian-drug regimen.
Outcomes and reported result
Primary endpoint: change in total Unified Parkinson's Disease Rating Scale score from baseline to week 48; additional UPDRS analyses, medication stability, tolerability and adverse events.
At week 48, total UPDRS change favored H₂ water: median −1.0 and mean −5.7±8.4 in the H₂ group versus median +4.5 and mean +4.1±9.2 with placebo (p<0.05). Six of nine H₂ participants improved and one was unchanged. The article states that additional analyses did not differ between groups. No participant changed levodopa or other antiparkinsonian medication; no adverse effect was attributed to H₂ water.
The author-uploaded full-text copy, allocation and attrition, water dose and concentration, primary and additional results, medication stability and adverse-event reporting were checked together with PubMed and the publisher record.
A reported association, difference or mechanism is not automatically a clinical benefit. Null findings, outcome type, study design and precision all matter.
Cautions and applicability
Very small pilot with 17 completers, one dropout, a single Japanese setting, large outcome variability and multiple analyses. The positive result was the total-UPDRS change; additional analyses were null. The trial predates a later multicenter randomized trial that did not confirm benefit. No dedicated funding or conflict-of-interest statement was identified in the checked article copy, so absence of funding or conflicts is not inferred.
Levodopa-treated Japanese Parkinson's patients under this 48-week pilot protocol; the result is hypothesis-generating, not definitive evidence of disease modification.
Cochrane RoB 2 · parallel randomized trial · headline week-48 total UPDRS result
Some concerns — preliminary; independent reviewer pending
Small pilot, limited reporting of allocation concealment, one participant absent from the week-48 analysis, and no completed comparison with a prespecified analysis plan.
Some concerns — randomization is reported, but sequence generation and concealment are not fully documented in the extracted report. · Some concerns — placebo preparation and double blinding are described; adherence and protocol-deviation detail is limited. · Some concerns — 17 of 18 randomized participants were included at week 48; the impact of the missing result requires review. · Some concerns — UPDRS is appropriate, but assessor-blinding procedures need fuller verification. · Some concerns — multiple outcomes and time points; protocol and prespecified analysis have not been independently compared with the report.
No single-study GRADE certainty rating is assigned. Certainty is assessed by important outcome across a complete eligible evidence set, not by attaching a final grade to one source.
Sources and record status
PMID: 23400965 · DOI: 10.1002/mds.25375
The publisher page is paywalled; a free author-uploaded full-text copy is available on ResearchGate.
Author-uploaded full-text PDF, PubMed record and publisher metadata; full-text extraction checked 9 August 2026.
2026-08-10