A major new analysis of more than 25,000 calls and text messages to a national psychedelic helpline offers what its researchers describe as the largest real‑world portrait to date of who seeks support during or after a psychoactive drug experience. The study, published in Science Advances, finds that distress rates are broadly similar across substances once the timing of the call is taken into account, and that most callers whose setting was recorded were alone. By linking thousands of anonymous support logs with caller feedback surveys, the dataset provides a new way to study whether individual peer supporters influence caller outcomes.
Psychedelics such as psilocybin and LSD can alter perception, mood and thought processes. When substances are used outside medical settings, peer support services offer non‑clinical aid from trained volunteers to calm a person during a difficult experience or to help with “integration” in the days or weeks after the event.
Although clinical trials of psilocybin and MDMA have attracted attention, most psychedelic use takes place in everyday environments—from private homes to music festivals. In the United States, public policy is shifting at state and local levels, with regulated frameworks in places like Oregon and Colorado and limited therapeutic access approved in others. Against this backdrop, health officials have lacked systematic data on what happens when non‑clinical experiences go awry and what kind of help people actually seek.
Bridging clinical trials and real‑world use
To fill that gap, Fireside Project founder Joshua White and colleagues built the Fireside Database, drawing on nearly five years of daily operations from the Psychedelic Support Line—a free national service running twelve hours daily via phone and text. The record comprises 25,419 peer‑support interactions handled by 220 trained volunteers, offering researchers a rare resource to study who seeks help outside clinical settings and how they rate the support they receive.
“Since our support line launched in April 2021, it has been building exactly that record, one conversation at a time,” White told PsyPost. “This paper turns it into a resource the whole field can use.”
Tracking five years of helpline activity
The researchers analysed 25,419 interactions between April 2021 and February 2026, of which 67.5% were voice calls and 32.5% were text messages, with 97.5% initiated by incoming callers. To protect privacy, the records automatically remove identifiers and separate phone numbers from conversations; callers could request deletion of anonymised recordings.
The database combines two layers of information. After each conversation, volunteers completed a standard log detailing duration, call type, substances reported, social setting, stage of the drug experience, emotional concerns and any shared mental‑health details. A follow‑up text the next day invited callers to complete a short satisfaction survey across five areas: whether they felt heard, overall quality, whether needs were met, whether they would call again, and whether they would recommend the helpline. A total of 2,777 surveys were completed, representing 10.9% of interactions.
What callers were seeking and how long it lasts
Across the period, about half of conversations ran six minutes or less, though the average length was close to 29 minutes due to longer discussions on occasion. The most frequent reason for contact was active support while under the influence (40.2%), followed by post‑experience integration (32.1%).
Other interactions fell outside the helpline’s remit, such as medical crises referred to emergency services or the 988 suicide and crisis lifeline (11.6%), general information requests (10.6%), or support for friends or trip sitters (smaller shares). Usage peaked on Saturdays, with roughly 18 interactions per day, and monthly highs were in August (479 interactions) versus a seasonal low in November (373).
Substance information was recorded for 15,596 interactions; the remaining 38.6% did not specify a drug. Among those with data, psilocybin was the most common (59.9%), followed by LSD (14.4%), cannabis (12.8%), ketamine (7.7%), MDMA (4.4%) and ayahuasca (3.2%). Reported substances also included DMT compounds, San Pedro cactus, Amanita muscaria and ibogaine. When multiple substances were reported, MDMA and LSD co‑use occurred more often than would be expected by chance.
Distress and timing—what matters most
In raw totals, distress among callers appeared highest for cannabis (47.2%), followed by psilocybin (42.2%), LSD (41.8%), ketamine (40.9%) and MDMA (40.7%). However, the authors note that cannabis calls were more often made while the caller was still intoxicated (70.2%), whereas most MDMA (66.3%) and ketamine (59.0%) calls occurred after the peak for integration.
When active‑trip calls are separated from integration calls, distress during active experiences clusters around 47% across major substances, while distress during integration sits around 36% (ranging from 32.3% for LSD to 38.8% for MDMA). In either phase, substance differences were not statistically significant, which the authors say points to the stage of the experience being a stronger predictor of distress than the drug itself.
“When people struggle with psychedelics, timing matters more than the substance,” White stated. He added that the helpline represents “real safety infrastructure, and it can be available to anyone, anywhere.”
He also noted that cannabis appearing at the top of distress rankings was surprising, but attributed it to timing, given that most cannabis callers were in the middle of the experience.
Who’s using the helpline and in what contexts
Context data from 6,998 interactions show that 66.3% of callers were completely alone, 26.4% were with one or a few close companions, and 5.0% were in ceremonial or larger group settings. Among 6,388 active‑trip interactions with timing known, 43.9% occurred after peak effects, 33.3% during peak and 22.8% as the drug was taking effect. Anxiety and feeling overwhelmed were the most common emotional concerns, while 911 or 988 escalations remained under 1% for all major substances—0.08% for ketamine and 0.65% for cannabis.
In the subset of 7,826 interactions where mental‑health history was recorded, 60.5% reported at least one psychiatric diagnosis. PTSD was the most common at 18.5%, followed by major depression (16.7%) and generalised anxiety disorder (8.8%). Notably, 2.4% (185 cases) reported schizophrenia, a group often excluded from clinical psychedelic trials. Demographic data show 46.8% male, 39.4% female and 13.8% nonbinary or another gender, with 68.4% identifying as white.
Allegations of facilitator abuse and responses
Forty‑three interactions (0.2%) involved callers reporting abuse by psychedelic facilitators or guides. Volunteers directed these callers to survivor peer‑support resources, without collecting identifying details about the alleged facilitators. In descriptive comparisons, abuse reports clustered around ceremonial settings and plant medicines such as ayahuasca, 5‑MeO‑DMT, San Pedro and iboga or ibogaine, accounting for 40.7% of listed substances (11 cases) versus 4.4% of substance‑coded interactions overall.
Reports of abuse within ceremonial contexts and plant medicine use also accounted for a large share of social‑context abuse cases. Among the 15 abuse reports with recorded emotional concerns, trauma processing was present in 10 cases (66.7%), compared with a 6.5% base rate across the dataset, and 34.8% of those with diagnosis data listed self‑reported PTSD, nearly double the overall rate.
White described the concentration of abuse reports in ceremonial and plant‑medicine contexts as troubling but noted that the data rely on callers’ accounts, not independent verification, and should be treated as a signal for further study.
Caller satisfaction and variation among supporters
Feedback from the 2,777 post‑call surveys was largely positive: 76.1% strongly agreed that they felt supported, 78.2% rated the quality of support as excellent, 66.8% said almost all needs were met, and more than eight in ten (86%) would definitely call again or recommend the service. Small minorities expressed dissatisfaction: 3.2% did not feel supported, and 2.1% rated the quality as poor. Among 30 negative written responses, the most common criticisms were an impersonal or judgmental tone (40%), a poor conversational fit (26.7%) and technical interruptions or delays (23.3%).
In traditional psychotherapy, therapist effects account for a small portion of outcome variance. The Fireside dataset includes 50 volunteers who completed at least 10 surveys each; researchers found an average satisfaction rating of 4.66 out of 5, with variation from one supporter to another. The authors emphasise that this is descriptive, not a causal test of whether individual supporters affect outcomes.
“This paper is best understood as a resource rather than a single finding,” White said. The researchers have posted the anonymised records and the code behind their analyses for other researchers to use. “Because 50 of our supporters each have 10 or more surveyed conversations, researchers can now ask a question the field has rarely been able to test at scale: does who supports you change how you fare?”
Joshua C. Black, a senior research scientist at Rocky Mountain Poison & Drug Safety, who was not involved in the study, welcomed the work. He said Fireside “meets a care need” and may escalate to 911/988 if required, and that the finding—distress rates not varying much by substance—provides valuable information for the public health field on how best to meet those in distress from psychedelic and cannabis experiences.
Limitations, cautions and future directions
The authors acknowledge several caveats. The study was funded by Fireside Project and conducted by its founder, its in‑house research arm and a paid consultant, so satisfaction results reflect an organisation evaluating its own service. Because the data come from a single helpline, the findings may not generalise to other peer‑support lines or harm‑reduction services, and because only satisfaction was measured, the study cannot determine whether the helpline prevented harm.
Moreover, the database captures only those who actively sought help, meaning distress rates reflect experiences of help‑seekers rather than the general population. Given cannabis’ wider prevalence, even a small adverse‑experience rate among users can generate a high volume of helpline contacts.
“Our distress rates describe our callers, not the population‑level risk of any drug, so please don’t read them as ‘47% of cannabis users have a bad trip,’” White stressed. Black also cautioned that these data reflect people in crisis moments and may not represent overall distress rates among all users. The authors acknowledge this limitation, common to call‑centre data.
Only 10.9% of interactions produced a completed follow‑up survey, so those with particularly positive or negative experiences may be overrepresented in the satisfaction results.
The team describe the current work as descriptive and plan future analyses using multilevel modelling to separate the influence of individual supporters from call difficulty, together with automated analysis of written feedback to identify which interpersonal skills best predict positive outcomes. White said: “The question I care most about is what makes support actually work: which skills and qualities in a supporter predict better outcomes for the person they’re helping.” He added that he hoped outside researchers would engage with the data and invited them to contact the project.
“Every data point in this paper is a real person at a vulnerable moment who trusted a stranger enough to reach out, and I’m deeply grateful to them and to the volunteers who answered,” White concluded. “If any of your readers ever need support during or after a psychedelic experience, our line is free and confidential. Download our free mobile app, or call or text 623-473-7433, every day from 11 a.m. to 11 p.m. Pacific.”
The study, “The Fireside Database: Large-scale evidence from 25,419 psychedelic support interactions,” was authored by Felix Schoeller, Marc Santolini, Dana Xiadani, Mickael Eskinazi, Ashwin Murali, David Esselman, Wyatt Houtz, Wesley Bellanca and Joshua White.
