Indication-stratified mortality risk of ibogaine treatment under contemporary safety protocols: a multisite analysis of 19,071 patients and updated systematic review of fatalities

Ibogaine-assisted therapy had already shown early promise for substance use disorders, including opioid use disorder, alcohol use disorder, and cocaine dependence. More recently, interest had expanded to non-substance-use indications, particularly traumatic brain injury with comorbid post-traumatic stress disorder, where earlier prospective work suggested substantial symptom improvement without serious cardiac events. At the same time, ibogaine’s safety remained a central concern because of recognised cardiac risks, including QTc prolongation and torsades de pointes, and older mortality estimates were based on data collected before contemporary safety protocols were introduced. The paper argues that the key unanswered question is not only overall risk, but whether fatality risk is concentrated in particular patient groups under current guideline-concordant care. Arns and colleagues therefore set out to examine ibogaine-associated mortality in two complementary ways: a multisite retrospective analysis of 19,071 treated patients from international clinics using established safety guidelines, and an updated systematic review of reported fatalities from the published and grey literature. Their main aim was to determine how mortality risk is distributed across treatment indications, especially whether patients treated for substance use disorders, and opioid detoxification in particular, face higher risk than those treated for non-SUD indications. The authors state that this evidence is intended to inform patient selection, cardiac risk mitigation, regulatory evaluation, and the design of forthcoming controlled trials.

The study was a retrospective observational analysis combining a multisite clinical cohort with an updated review of reported fatalities. The clinical component drew on international clinics that administered ibogaine according to the Clinical Guidelines for Ibogaine-Assisted Detoxification. Clinics were identified through professional networks and direct outreach; 16 were invited and 11 contributed data. Data were aggregated and anonymised, with no individual-level identifiable information collected. Clinics were assessed using the IBO-SAFE checklist, a structured safety evaluation covering adherence to published guidelines such as pre-treatment screening, cardiac monitoring, and medical oversight. Participating clinics supplied aggregate information for 2016-2026, or longer where available, including the number of treated patients, deaths occurring within 72 hours of ibogaine administration, the proportion treated for substance use disorder versus non-SUD indications, and summary serious adverse event information. Fatalities were reported using clinic records and case review, but the paper states that deaths were not independently adjudicated. The primary outcome was death during or within 72 hours of ibogaine administration. Indication was classified at the clinic level as SUD or non-SUD. Seizures were a secondary outcome, reported separately in the supplementary material. For the literature component, the authors searched PubMed using terms including ibogaine, iboga, fatality, and death, and also conducted structured grey-literature searches in legal databases and media reports, including non-English material. The final search date was 9 April 2026. Cases were included if death occurred within 72 hours of reported ibogaine exposure, regardless of causal attribution, and were cross-referenced and de-duplicated against earlier inventories and a research database. Statistical analysis used mortality rates calculated as deaths divided by treated patients. Group comparisons between SUD and non-SUD were tested with Fisher’s exact test; when cells were zero, a Haldane-Anscombe correction was applied to estimate odds ratios. Temporal trends in fatalities were assessed with Spearman correlation against Google Trends search interest, and pre- versus post-guideline fatality rates were compared using a one-tailed t-test.

In the multisite clinical cohort, 11 of 16 invited clinics participated, giving a 69% response rate. The participating sites showed high adherence to safety guidelines, with IBO-SAFE scores ranging from 7.3 to 9.8 and a mean of 8.8/10. Across 19,071 patients treated between 1996 and 2026, six deaths occurred within 72 hours of ibogaine administration, corresponding to a mortality rate of 0.03% or 0.3 per 1,000 patients. All six deaths occurred among patients treated for substance use disorder, specifically opioid use disorder: 6 of 10,382 OUD patients died, whereas no deaths were reported among 8,689 patients treated for non-SUD indications. This difference was statistically significant (Fisher’s exact test, P=0.026). Because there were no deaths in the non-SUD group, the odds ratio could not be estimated directly; with a Haldane-Anscombe correction, the estimated odds ratio was 10.9, but the confidence interval was very wide (0.61 to 193), showing that the estimate was imprecise. All clinical-cohort deaths occurred in males aged 45-49, and where doses were available they were 14-15 mg/kg. The reported substances prompting treatment included fentanyl, heroin, methadone and oxycodone, and unknown opioids. Reported causes of death were consistent with cardiotoxicity. No deaths were seen in the non-SUD group. The updated literature and grey-literature review identified 16 additional fatality cases, bringing the total to 48 reported deaths between 1990 and 2026. Fatalities were reported in 15 countries, most often Mexico, the United States, the United Kingdom, the Netherlands, and Costa Rica. Among cases with available data, the mean age was 39.4 years and most were male. Of the 45 cases with a reported indication, 42 involved SUD. Most fatalities occurred in clinical settings, followed by home or unsupervised settings, underground or unregulated settings, and ceremonial or ritual contexts. Pre-treatment ECG screening was documented in only 10 cases, mostly after 2016. Where causes of death were available, the most common categories were cardiac or cardiovascular events and drug intoxication or combined toxicity. Polysubstance use was common, and post-mortem toxicology often detected opioids, benzodiazepines, and cocaine. Three deaths from the clinical cohort were also found in grey literature and were de-duplicated. Temporal analyses showed that annual fatality counts from 2004-2026 were moderately correlated with public interest as measured by Google Trends overall (Spearman r=0.47, P=0.026), with a stronger correlation in the pre-guideline period (r=0.59, P=0.036) and no significant association after guideline implementation (r=0.35, P=0.32). Mean annual fatalities were higher before guidelines than after them (2.15 versus 1.1; P=0.04). In the combined dataset, 41 of 44 deaths with known indication occurred in people treated for SUD, a distribution highly unlikely under a 50% null assumption (binomial P=1.6×10⁻⁹). All three non-SUD deaths occurred before 2016 and in non-clinical or ceremonial settings. The sensitivity analysis suggested that, to raise the combined mortality rate to the historical estimate of 0.23%, the five non-participating clinics would need to have had roughly 42 to 52 unobserved deaths, implying mortality rates of about 0.9% to 2.1% at those sites.

The authors interpret the findings as showing that ibogaine-associated mortality is concentrated in patients undergoing opioid detoxification, rather than being spread evenly across all treatment indications. They emphasise that, in the clinical cohort, six deaths among 19,071 treatments were all confined to OUD patients, and that the literature showed the same pattern, with 93% of fatalities with known indication occurring in SUD contexts. On this basis, they conclude that fatality risk appears low overall under contemporary guideline-concordant care, but clusters in the opioid-detoxification population. Arns and colleagues stress that the 0.03% mortality rate observed in the clinical cohort should be treated as a lower bound rather than a precise point estimate, because participation was voluntary, the clinics contributed aggregated data under non-disclosure agreements, and deaths were self-reported without independent adjudication. They note that these features could bias the absolute rate downward. However, they argue that the indication-stratified pattern is more robust to these limitations. They also state that the sensitivity analysis makes it unlikely that enough unobserved deaths at non-participating sites would exist to overturn the general conclusion of lower current mortality than historical estimates. Relative to earlier research, the authors argue that the present results are consistent with the idea that structured safety protocols reduce mortality. Historical estimates were derived from periods before standardised screening, monitoring, and exclusion criteria were widely implemented, and the study’s lower observed mortality aligned with high protocol adherence. In the broader literature, reported fatalities appeared to track public interest more closely before the guidelines than afterwards, which the authors see as compatible with improved safety practices, though they explicitly say this does not prove causality. They also note that opioid detoxification may carry special risk because of withdrawal-associated electrolyte disturbance, autonomic instability, and polysubstance exposure, with fentanyl exposure highlighted as a possible contributor, though the sample is too small for firm conclusions. For non-SUD indications, they point out that the three deaths found in the literature occurred before the 2016 safety guidelines and in non-clinical or ceremonial settings, often with pre-existing cardiac disease. The authors acknowledge several limitations: the retrospective design, aggregated self-reported clinic data, possible under-ascertainment of deaths outside participating clinics or in unregulated settings, likely undercounting of recent fatalities in the literature, heterogeneity introduced by grey-literature sources, and the focus on acute 72-hour outcomes rather than longer-term risk. They also could not systematically evaluate variation in formulation or dose. At the same time, they highlight strengths including the large scale of the clinical cohort, the standardised assessment of protocol adherence, and the convergence between clinical and literature-based data sources. They suggest that the IBO-SAFE assessment tool may help Institutional Review Boards and investigators evaluate safety procedures in future trials, and they argue that current results provide important population-level safety context for forthcoming ibogaine and noribogaine studies, especially those involving patients with OUD and cardiac risk factors.