Ozone Therapy Adverse Events by Route: What Case Reports Show, and Where EBO2 (EBOO) Fits

We could not find a published case report of a serious adverse event during EBO2 (also called EBOO), as of October 2, 2026. The adverse events in the medical literature come from other ways of giving ozone: mixing it into a bag of blood and returning the blood, injecting the gas into a vein, and injecting it into or beside the spine, a joint, a wound, or the skin. This page lists every such report we could read, grouped by route, with what each says happened, the outcome, and how its authors explain it. It ends with how to search FDA’s adverse-event database and how to report a problem to FDA.

Ozone therapy is not FDA-approved to treat any condition, and federal regulation calls ozone “a toxic gas with no known useful medical application in specific, adjunctive, or preventive therapy” [55]. For what the EBOO studies record and the warning signs after a session, see our side effects and safety guide; for who the sources say should not be treated, see contraindications.

What a case report can and cannot show

A case report describes one patient, or a few, as the treating doctors saw them. It shows that an event followed a procedure and how those doctors explained it. It cannot show how often the event happens, who is most at risk, or on its own that ozone caused it. Authors word causation differently, from “most likely” to “probable” to a plain statement that a report establishes only a link in time, and we keep their wording below.

No registry counts how many ozone treatments are given or what follows them. A 2026 scoping review of autohemotherapy safety concluded that the overall rate of serious adverse events cannot be reliably estimated, because there are no prospective registries and no reliable figures for how many treatments are given [1]. A 2026 umbrella review of randomized trials found safety outcomes too inconsistently reported to support firm conclusions [2]. Velio Bocci, a Siena physiologist whose group reported about 800 EBOO sessions, wrote that he had proposed six times that health authorities require ozone therapists to keep a register of adverse events [3].

The low complication rates quoted in practitioner reviews, 0.0007% or 0.7 per 100,000 treatments, trace back to a 1982 survey by Jacobs [3][4]. Bocci notes that the survey’s data included four deaths from direct intravenous injection of the gas [3], and the 2026 scoping review says the frequently cited low estimate rests on uncertain methods [1].

Blood-based methods

These reports involve ozone mixed into withdrawn blood that is then returned (autohemotherapy), or gas given into a vein. Reports marked “abstract only” are ones whose full text we could not read.

Milder effects appear in Bocci’s account of about 5,000 major autohemotherapy treatments in Siena from 1995 to 2000: tingling of the lips and tongue near the end of reinfusion, which he attributes to excess citrate; nausea, bloating, and a metallic taste in 10 to 15% of women, which he attributes to additives in PVC bags; a day of tiredness after the first treatments in 20 to 30% of patients; and four women who developed a rash, itching, nausea, flushing, and slightly low blood pressure. He writes that these stopped after his group switched to glass bottles and a calibrated amount of citrate [3]. A 2025 modeling paper by authors from the Italian society SIOOT argues the opposite about containers, that glass bottles cause more breakdown of red blood cells than plastic bags [56].

Three more reports in this group are letters we could read only by title: hepatitis C transmission by ozone enrichment of blood (1996) [8], hepatitis C and HIV infection after autohemotherapy (1997) [9], and methemoglobinemia after autohemotherapy given with high-dose vitamin C (2022) [15].

Injecting the gas itself into a vein is a different practice from autohemotherapy. Bocci’s 2004 review notes four recorded deaths from pulmonary embolism during direct intravenous administration [17] and wrote that the gas mixture should never be injected into blood vessels because of the risk of oxygen embolism [16]. A 2018 document from Tuscany’s regional clinical-governance body summarizes Italian ozone-society guidelines as saying the mixture “must not be used in direct intravenous administration” because of the danger of pulmonary gas embolism (our translation) [18].

Three official records describe harm after intravenous ozone. In New York, a medical examiner found gas emboli in a podiatrist’s patient who went into cardiac arrest within 15 minutes of an intravenous ozone injection, attributed them to the ozone, and ruled the death a homicide due to extreme medical negligence; an appellate court held that determination was not arbitrary and capricious [19]. In California, a physician admitted an accusation that a patient’s IV made gurgling sounds during intravenous ozone therapy in 2020, after which she lost consciousness and was diagnosed with air embolism and a stroke; he was placed on probation and barred from intravenous ozone therapy [20]. In New South Wales, the Health Care Complaints Commission found, on expert opinion, that treatment at an ozone clinic was the most likely reason a patient developed an infection that led to septic shock [21]. Our side effects guide gives the exact wording of each.

Injections into or beside the spine

These reports involve injections into a disc (intradiscal), around the nerve root (periganglionic or intraforaminal), into the back muscles (paravertebral), or into the epidural space, almost all for back or neck pain.

Bocci’s chapter also records a death in Naples in May 2001 that he attributes to intramuscular injection of the gas into the muscles beside the spine, and it describes how a large, painful injection can trigger a reflex slowing of the heart that may end in cardiac arrest [3].

Larger summaries put these reports in context without settling the rate. A 2010 meta-analysis of 12 studies and almost 8,000 patients gave a complication likelihood of 0.064% for disc treatment; its authors’ listed affiliation is ActiveO, of Salt Lake City [38]. A 2025 analysis by authors from the Italian ozone society SIOOT, pooling seven studies of 120 patients, reported a relative risk of severe adverse events 6.57 times higher for intradiscal than for intramuscular injection [39].

Other routes

Two more are known to us only by title: posterior reversible encephalopathy syndrome after intramuscular injection (2020) [44] and massive emphysema with air in the chest after ozone therapy (2021) [45]. Breathing ozone is a separate hazard: FDA has received reports of asthma attacks, headaches, and breathlessness from people who used ozone devices sold to clean CPAP equipment [46].

What the reports have in common

• Timing. Most neurological and heart events began during the procedure or within minutes of it [13][25][29][30][40].
• Gas in the circulation. Many authors attribute the event to gas embolism. In five reports, echocardiography found a patent foramen ovale, an opening between the heart’s upper chambers, which the authors describe as the route by which bubbles crossed from veins to arteries [13][25][28][30][31]. A case report cannot show how much such an opening adds to the risk.
• Infection. Infections followed blood procedures and injections alike, and several authors tie them to lapses in sterile technique [7][32][34][36]. The New South Wales commission found a clinic trying to sterilize single-use items for reuse [21].
• Errors and equipment. One arrest followed an injected volume more than twice the planned one [29], and the California accusation charged a failure to maintain the ozone machine [20].

Practitioners read the same reports differently. A 2020 review in a journal special issue on ozone therapy concluded that most safety problems in the case reports come from infections or traumatic injuries due to malpractice, not from ozone itself at therapeutic doses [57]. The 2026 scoping review, by contrast, groups the documented events by mechanism, including gas embolism, hemolysis at high concentrations, and reactions to equipment materials, and says their overall frequency cannot be estimated [1].

What has not been reported: EBO2 itself

A Europe PMC search on October 2, 2026, for “extracorporeal blood oxygenation and ozonation” or EBOO returned 88 records; none of the EBOO papers among them is an adverse-event report [50]. The largest controlled EBOO trial recorded no side effects or complications in 210 treatments [47], and a 2005 review by the same Siena group reports more than 1,200 treatments in 82 patients and describes treating up to 4,800 mL of heparinized blood in an hour without technical or clinical problems [48]. A California clinic’s paper on ozone dialysis, which it says is commonly referred to as EBOO, reports at least 400 sessions with no untoward effects observed [49]. These come from the people who developed or offer the procedure, none was designed to detect rare events, and no registry exists. The EBOO studies side by side show how small they are.

How to search FDA’s MAUDE database

MAUDE holds medical device reports sent to FDA by manufacturers, importers, and hospitals and other user facilities, which must report, and by clinicians, patients, and consumers, who may [51].

1. Open the simple search at https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfmaude/TextSearch.cfm. FDA’s instructions say to enter a single word, an exact phrase in quotes, or several words joined by “and” [51].
2. Choose a report year, or the option for all years. MAUDE covers the last ten years and is updated monthly; older reports are in FDA’s MDR data files [5].
3. Try terms such as ozone, "ozone therapy", the brand of an ozone generator, or the name of a device a clinic says it uses; our devices guide explains how to look a device up in FDA’s other databases.
4. For more control, use the advanced search at https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfmaude/search.cfm to filter by brand name, manufacturer, product code, event type, or date; each search returns up to 500 reports [51].
5. Open a report and read the event description, the event type (death, injury, or malfunction), who reported it, and the date FDA received it.

What a result means: FDA says a report does not necessarily show that the device caused or contributed to the event, and that the report data “is not intended to be used either to evaluate rates of adverse events” or to compare devices; under-reporting, inaccuracies, and unverified details limit it [5]. A search that finds nothing does not mean nothing happened.

We ran these searches on October 2, 2026, through openFDA, FDA’s public interface to the same reports (data last updated September 22, 2026). The phrase “ozone therapy” appeared in the narratives of 10 reports [52]. Eight concerned other products, such as breast implants, dermal fillers, a needle, and a spinal stimulator, with ozone therapy mentioned in passing. Two were voluntary reports about ozone treatment itself: one described loss of consciousness and days in a coma after ozone therapy, and one, coded as a death, described ozone treatment of the blood. Like any MAUDE report, neither establishes cause [5]. The term EBOO returned no reports.

How to report a problem to FDA

1. Use MedWatch, FDA’s voluntary reporting program. The online form is at https://www.accessdata.fda.gov/scripts/medwatch/index.cfm, with one version for health professionals (Form 3500) and one for consumers and patients (Form 3500B) [54].
2. PDF versions are at https://www.fda.gov/media/76299/download (Form 3500) and https://www.fda.gov/media/85598/download (Form 3500B). FDA also takes reports by phone at 1-888-INFO-FDA (1-888-463-6332), option 2 [53].
3. FDA encourages people to take the form to their doctor, who can add clinical details from the medical record; a health care provider is not required to report, and patients may file themselves [53].
4. The online form can be saved and finished within 3 days, contact details are optional, and a reporter can ask FDA not to share them with the manufacturer [54].
5. Include the product and a summary of what happened [54]. The details most often missing from the published reports above are the route, the ozone concentration and gas volume, any anticoagulant used (see our blood thinners and medicines guide), the generator or device name, and how soon symptoms began.

The form’s page tells anyone with a medical emergency to call 911 [54]. MedWatch collects reports about medical products [53]; conduct by a practitioner is a matter for state licensing boards, such as the Medical Board of California in the case above [20], and our regulatory tracker lists the records we hold. Our clinic directory lists what each clinic states about its own service.

What we could not verify

• Five reports we know only by title: the 1996 and 1997 hepatitis letters [8][9], the 2022 methemoglobinemia letter [15], the 2020 encephalopathy letter [44], and the 2021 emphysema report [45]. We list them so readers can find them, not as evidence of what happened.
• Twelve reports read as abstracts only, marked in the tables. Their full texts may give doses, timing, or other causes we could not see. We also cite the 2010 disc meta-analysis and the 2026 scoping review from their abstracts [1][38].
• The 1982 Jacobs survey behind the 0.0007% figure, which we have not read.
• Doses. Most reports give no ozone concentration or gas volume, so no pattern by dose can be drawn.
• Unpublished events. Mild reactions, and serious ones nobody wrote up, do not appear in journals, and MAUDE captures few ozone treatments.
• Comment letters that journals published about several of these reports, which we did not read.

How this guide was made

We built this compilation from 57 sources: case reports, series, and reviews read through NCBI and Europe PMC, court and regulator records, and FDA’s own pages and data. The study details come from our library’s study cards where one exists (see our research library) and otherwise from the papers’ abstracts or full texts as marked. The MAUDE counts come from our own openFDA queries on October 2, 2026, which anyone can repeat with the links in the sources. Drafting used AI tools. A human editor has not yet checked this guide claim by claim against its sources, and no clinical reviewer has signed off yet.