Psychedelic Safety Outside Clinical Trials: What a National U.S. Survey Found

Clinical trials can tell researchers a great deal about how a substance behaves under controlled conditions. Participants may be screened for medical and psychiatric risks, doses can be measured, substances can be tested for purity, and people may receive professional monitoring and support.

Real-world use can look very different.

A study published September 29, 2026, in JAMA Network Open examined health outcomes that U.S. adults attributed to recent use of psilocybin, LSD, MDMA, ibogaine, and 5-MeO-DMT. Rather than testing these substances experimentally, researchers asked people about their experiences and used statistical weighting to estimate how commonly different outcomes may occur among U.S. adults reporting use.

The findings provide useful information about psychedelic safety outside clinical trials. They also require careful interpretation. The study can tell us what people reported and associated with their substance use. It cannot establish that a particular substance caused those outcomes.

Research involving psychedelics and related substances has increasingly examined possible therapeutic benefits. But results from clinical studies do not necessarily describe what happens when these substances are used in the broader population.

Outside research settings, doses and purity may vary. People may use multiple substances, and medical supervision may be absent. The researchers wanted to better understand health outcomes associated with recent real-world use of five substances:

Their goal was descriptive. They were looking for patterns in outcomes that people themselves attributed to these substances, rather than attempting to determine whether one substance directly caused a particular health effect.

The researchers used data from the 2025 National Survey Investigating Hallucinogenic Trends, or NSIHT. This was a double opt-in online survey involving adults across all 50 U.S. states. Responses were statistically weighted against a nationally representative survey using a data-fusion and calibration method designed to improve estimates for the U.S. adult population.

Among 1,964 survey respondents, past-year use was reported by:

The weighted estimates suggested that past-year use was much more common for some substances than others. Researchers estimated approximately 8.12 million U.S. adults had used psilocybin during the previous year, compared with 3.93 million for MDMA, 2.58 million for LSD, 740,000 for 5-MeO-DMT, and 560,000 for ibogaine.

People reporting use within the previous 12 months were asked about changes in physical health, mental health, and use of health care services that they attributed to each substance.

Those reporting use within the previous three months were also asked whether they had experienced any of 13 selected adverse effects during that period and attributed them to the substance. The researchers used a shorter three-month window for these symptoms in part to reduce problems with remembering events over longer periods.

Importantly, these were self-attributed outcomes. Researchers did not independently determine whether the substance caused each reported symptom.

Two different ways of looking at the results tell different parts of the story.

Among recent users, the proportion of people reporting adverse effects was generally highest for ibogaine and 5-MeO-DMT. For every one of the 13 adverse effects examined, the highest percentages occurred among respondents reporting use of one of these two substances.

When researchers looked at whether participants reported at least one of the 13 adverse effects, the estimates were:

These percentages should not be interpreted as the percentage of users experiencing a serious medical emergency. The category included 13 different self-reported symptoms, and reporting any one of them placed someone in this group.

At the same time, psilocybin, LSD, and MDMA were used by far more people. As a result, the estimated total number of U.S. adults experiencing particular outcomes was often greater for these more widely used substances, even when the percentage of users reporting an outcome was lower.

This distinction matters. A relatively uncommon substance can have a high proportion of users reporting an outcome while still affecting fewer people nationally. A much more commonly used substance can have a lower proportion reporting that same outcome but represent a larger number of people overall.

The study also asked about perceived improvements, not only adverse outcomes.

Among respondents reporting psilocybin use within the previous year, 51.2% reported a mental health improvement that they attributed to psilocybin, while 4.7% reported mental health worsening. For physical health, 24.0% reported improvement and 6.2% reported worsening.

These findings are worth reading carefully.

They do not demonstrate that psilocybin improved mental health in 51.2% of people who used it. Participants were reporting their own perceptions, and there was no untreated or placebo comparison group. People who chose to use psilocybin may also differ from people who did not use it in ways that affect these results.

What the study does show is that perceived benefit and perceived harm can exist within the same real-world dataset. That makes safety research more complicated than simply asking whether psychedelics are “helpful” or “harmful.”

One finding deserves particular attention because it is consistent with concerns identified in previous research.

Among respondents reporting recent ibogaine use, 44.2% attributed heart palpitations, chest pain, or chest pressure to ibogaine. The 95% confidence interval ranged from 29.2% to 59.2%. Among recent 5-MeO-DMT users, 51.3% reported this category of symptoms, with a confidence interval from 37.7% to 64.9%.

The ibogaine finding is relevant because previous research has identified an association between ibogaine and QTc prolongation, a change in the heart's electrical activity that can increase the risk of dangerous abnormal heart rhythms in some circumstances. The authors specifically noted that the cardiac symptoms reported in their survey are consistent with known concerns about QTc prolongation and arrhythmia associated with ibogaine.

That does not mean this survey established that ibogaine caused the cardiac symptoms reported by individual participants. It does mean the pattern aligns with an existing area of medical concern and supports continued attention to cardiovascular safety.

The study's main figure presents the same data from two different perspectives.

Panel A asks: Among people who recently used this substance, what percentage reported this adverse effect?

This makes it easier to compare the proportion of users reporting an outcome. Ibogaine and 5-MeO-DMT stand out in this panel because their percentages were generally higher.

Panel B asks: About how many U.S. adults could that percentage represent?

This takes the estimated prevalence of substance use into account. Because psilocybin, LSD, and MDMA were used by more people, their national estimates can be larger even when the proportion of individual users reporting an effect was lower.

The horizontal lines around the estimates represent 95% confidence intervals. They show the range of values compatible with the survey estimate under the statistical method used.

Wider intervals mean greater uncertainty.

This is particularly important for ibogaine and 5-MeO-DMT. Only 59 respondents reported past-three-month ibogaine use and 72 reported past-three-month 5-MeO-DMT use. Because those groups were small, many of their estimates have wide confidence intervals. The apparent differences between substances therefore should not be treated as precise measurements of comparative risk.

The study has several important limitations.

First, it was cross-sectional. Researchers collected information about substance use and reported health outcomes rather than assigning substances and following participants under controlled conditions. This means the study cannot establish cause and effect.

Substance co-use was also common. Depending on the group, approximately 66% to 78% of respondents reported some form of substance co-use. An outcome attributed by a participant to one substance could therefore have been influenced by another substance or an interaction between substances.

The groups were not identical before taking the substances. They differed in psychiatric history, patterns of use, therapeutic or clinical use, and other characteristics. For example, lifetime histories of schizophrenia and psychosis differed across substance groups. These differences make it difficult to determine how much of the observed pattern reflects the pharmacology of a particular substance versus characteristics of the people using it.

Real-world use also means researchers were not standardizing dose, purity, setting, frequency, or supervision as they might in a clinical trial.

Finally, the study depended on memory and self-attribution. Someone may reasonably believe that a symptom was related to a substance without that relationship being medically confirmed. The authors also note that the nature of an experience itself may influence attribution. A prolonged or physically demanding ibogaine experience, for example, might make someone more likely to connect physical symptoms with the substance.

Clinical trials are important for understanding potential therapeutic effects and risks under controlled conditions. Population-level research adds a different kind of information: what people report when substances are used in a much wider range of circumstances.

That difference has practical implications for harm reduction.

Information about potential risks can help people make more informed decisions about medical screening, medication and substance interactions, preparation, setting, and when medical supervision may be particularly important. When substances come from unregulated sources, drug checking may provide information about identity or some contaminants where appropriate and available, although testing cannot make substance use risk-free.

The ibogaine findings also reinforce why substance-specific risks matter. A general category such as “psychedelics” can hide meaningful differences in pharmacology and medical risk.

At the same time, this study did not test whether preparation, supervision, drug checking, screening, or any other harm-reduction practice prevented adverse outcomes. Those practices should therefore not be presented as conclusions demonstrated by this survey.

As psychedelic research and access expand, understanding potential benefits requires understanding potential harms as well.

Controlled trials answer important questions, but they usually involve selected participants, known substances, defined protocols, and structured monitoring. Real-world surveillance can help identify patterns that may not be as visible under those conditions.

This study adds a large U.S. population perspective. It suggests that the safety picture varies across substances and depends partly on how risk is measured. Ibogaine and 5-MeO-DMT showed higher proportions of users reporting adverse outcomes, while more commonly used substances such as psilocybin, LSD, and MDMA can account for larger numbers of affected people at the population level.

The same dataset also contained reports of perceived improvement. Most notably, more than half of respondents reporting past-year psilocybin use attributed some mental health improvement to it. Those reports are not proof of therapeutic effectiveness, just as self-reported adverse outcomes are not proof of causation. Together, they show why real-world psychedelic research needs to track both sides of the picture.

A nationally representative U.S. survey found that people reporting recent ibogaine or 5-MeO-DMT use had the highest proportions of self-attributed adverse effects among the five substances studied. Because psilocybin, LSD, and MDMA were much more widely used, however, they often represented larger estimated numbers of U.S. adults experiencing these outcomes.

The study identifies patterns worth investigating. It does not establish that any substance caused an individual health outcome, nor does it provide a direct ranking of which substance is safest or most dangerous.

As psychedelic use expands beyond research settings, continued safety monitoring can help build a clearer picture of both potential benefits and potential harms.

Adverse Health Outcomes of Psilocybin, LSD, MDMA, Ibogaine, and 5-MeO-DMT | JAMA Network Open