Retreatment Is the Hidden Access Question

The previous article in this series asked how long the benefit of psychedelic therapy lasts. That question matters because most of the cost arrives around treatment, while the health gain accumulates only for as long as someone remains better than they otherwise would have been.

But even a good answer about durability would leave us with another decision. What happens when someone does not improve enough? What happens when they improve for six months and then become unwell again? Do they receive another psychedelic session, repeat the full therapy programme, add another treatment, or wait?

The public story has often jumped over this part. Psychedelic treatment is contrasted with an antidepressant taken every day or therapy delivered every week. One or two profound sessions appear to replace continuing care. Robin Carhart-Harris has warned against the image that you can “click your fingers, one and done, miracle cure”. The phrase is memorable because it captures both the appeal and the risk of the story.

Saying that treatment may happen again does not make the first treatment a failure. Depression, PTSD, OCD, and substance-use disorders can be recurrent. A person can improve meaningfully, regain months of ordinary life, and still need care later. That is familiar in the rest of mental health care. It only feels awkward here because long-lasting change after limited treatment has become such a central part of the psychedelic promise.

Patient testimony makes the gap harder to ignore. At a public meeting on PTSD treatments, Karen Dunn described her own experience as “not a one and done situation”. She spoke about multiple sessions and skilled therapy. Her experience cannot establish a treatment schedule, but it is a useful correction to the idea that one administration is the whole intervention.

Retreatment is therefore not a footnote to durability. It is the bridge between a trial result and a care pathway. It asks who comes back, why they come back, what happens during the second episode, and whether the health system has kept enough money, staff, and rooms available for them.

The language becomes confusing quickly. A paper may refer to repeated doses, another to readministration, and a company to episodic treatment. Those terms can describe very different moments in care. This is therefore my attempt to separate five things that are often bundled together.

The easiest distinction is between completing the first treatment and starting another one. Three MDMA sessions planned from the beginning belong to one initial course. A second psilocybin administration offered because symptoms remain high may be an adaptive part of that course. Neither is the same as someone first recovering, relapsing nine months later, and returning for a new episode.

Maintenance is different again. Here the next treatment is scheduled before a full relapse, perhaps every few months, because regular treatment is expected to preserve the benefit. Classic psychedelic therapy does not yet have an established maintenance model comparable with esketamine, continuation ECT, or maintenance rTMS.

These categories are not just vocabulary. Each one needs a different trial. A planned second dose can show whether two administrations work better than one. It cannot show that someone who relapses after a year will respond again. A protocol that allows retreatment can show that researchers have made space for the question. It cannot show retreatment works until the results report who returned and what happened next.

For this article, I use retreatment in the narrower sense: someone has completed the initial course, initially benefited, later worsened or relapsed, and then receives another administration or treatment course. When the evidence concerns something else, I will call it something else.

Once the categories are separated, the evidence looks less mature than the phrase “repeat dosing” suggests. Psychedelic studies have accumulated useful experience with more than one supervised administration. Most of that experience still concerns the initial treatment course.

The pivotal MDMA-assisted therapy study in severe PTSD planned three long administration sessions about a month apart, alongside preparation and integration. That is strong evidence about a multi-session treatment package. It is not an answer to what should happen after someone later relapses. The US Food and Drug Administration made this gap explicit in its 2024 complete response letter, saying the application had not established whether one cycle was adequate for a chronic disorder or whether retreatment was necessary.

The EPIsoDE study in treatment-resistant depression also built two psilocybin administrations into the initial programme. The second session occurred six weeks after the first. The study did not observe a clear additional benefit from a second 25 mg administration within its timeframe. That does not prove a second administration is never useful. It does remind us that “more treatment” cannot simply be assumed to add another equal block of benefit.

Other trials are deliberately making the course more adaptive. In the DosOp study in US veterans with PTSD, participants can receive between one and five MDMA-assisted therapy cycles. After each cycle, the participant and therapists consider symptom change, tolerability, and burden before continuing. A Washington University psilocybin study plans to offer one optional readministration when depression symptoms remain above a protocol threshold at day 30.

These studies are asking a valuable question: can the number of sessions be learned from the person's early response, instead of being fixed for everyone? They may eventually help separate people who need no more treatment from those who might benefit from completing a longer initial course.

True relapse-triggered retreatment is harder to find. The registry for a Canadian psilocybin trial in treatment-resistant depression allows participants who relapse to receive up to two repeat doses. Yet the public evidence does not currently tell us the proportion who relapsed, how relapse was applied in practice, how many returned, or how well they responded after returning.

Newer programmes are trying to close that gap. The DT120 Ascend Phase III trial, led by Definium Therapeutics (formerly MindMed), includes a 40-week open-label extension in which treatment may be offered according to prespecified symptom and safety criteria. The COMP360 depression programme also includes longer observation and open-label treatment periods. Those designs show that retreatment has become a serious development question. Until their outcomes are published, they remain plans for producing evidence rather than the evidence itself.

The strongest conclusion is therefore also the narrowest. Multiple supervised administrations can form a feasible treatment course, and researchers are beginning to test response-based decisions. We still have very little published evidence showing that a person who benefits, later relapses, and receives another full psychedelic treatment reliably recovers again.

A health service cannot create one rule called “retreatment” and apply it to everyone. The next decision depends on the path that came before it.

A person who did not respond to the first administration is not in the same position as someone who recovered for a year and relapsed. Repeating the same intervention for a nonresponder might help if the first dose was insufficient or the person needed a longer initial course. It might also repeat an ineffective and burdensome experience. We do not yet have a validated rule for choosing between repetition, dose change, another therapy, or stopping.

Partial response raises a different possibility. Another administration might deepen an incomplete improvement. This is where adaptive studies such as DosOp are especially useful. But even here, services will need a threshold. A change that is statistically visible is not automatically enough to justify another all-day session and its surrounding care.

Later relapse is the clearest case for true retreatment. The first course has already shown that the person can benefit. That makes it plausible that another course could help, but plausibility is not a response rate. We need to know whether the second response is as likely, whether it lasts as long, and whether adverse events or treatment burden accumulate.

These decisions also require a stopping rule. A system that approves indefinite repeat treatment whenever symptoms return may expose patients to repeated burden without enough evidence. A system that permits only one lifetime course may deny a useful treatment to someone who responded well and later became ill again. The fair rule will probably depend on observed benefit, time since treatment, severity, other available care, safety, and patient preference.

Monitoring is what connects those factors. A retreatment pathway needs scheduled follow-up long enough to see deterioration, clear measures that can trigger a review, and a clinician who can distinguish relapse from a difficult week. Without that infrastructure, treatment is not episodic care. It is a series of disconnected episodes.

Psychedelic therapy is not the first mental health intervention to confront recurrence. Other fields have built continuation, booster, and maintenance pathways around treatments that can work well without working forever.

Psychotherapy provides the most intuitive comparison. CBT is usually delivered as a defined course, not as weekly therapy for life. People may later use booster sessions, a relapse-prevention plan, or another course when symptoms return. A meta-analysis of CBT and related approaches found evidence that they can reduce relapse and recurrence in depression. Another review of booster sessions reported a lower relapse risk when boosters were included.

The comparison helps with framing. Returning for psychotherapy does not erase the value of the first course. It can be normal management of a recurrent condition. Yet a CBT booster is not a direct template for psychedelic retreatment. It does not usually require drug screening, an all-day altered state, continuous monitoring, or a dedicated room.

Esketamine shows a more medicalised maintenance model. The US prescribing information for Spravato starts with twice-weekly treatment, moves to weekly treatment, and then recommends weekly or every-two-week maintenance at the least frequent schedule that preserves response. The treatment burden is substantial, but the pathway tells clinicians and payers when continuing care is expected.

Continuation ECT and maintenance rTMS make the same conceptual point. A strong acute response can be followed by scheduled treatment intended to prevent relapse. This is not necessarily evidence of failure. It is acknowledgement that the underlying illness can return.

What transfers to psychedelic therapy is the need for monitoring, thresholds, stopping rules, and an explicit plan for recurrence. What does not transfer cleanly is the delivery model. Repeating a psychedelic treatment may mean another medicine administration, preparation, integration, room, trained staff, transport plan, and day away from ordinary responsibilities. The return pathway has to be designed around that whole package.

The episode-cost article counted the resources surrounding one psychedelic treatment: screening, preparation, the administration day, staff, room time, monitoring, and integration. Retreatment reopens some or all of that bundle.

An economic model therefore needs at least three additional inputs. It needs the probability that someone returns, the timing of that return, and the cost of the repeat episode. It also needs a fourth input that is often missing: what health benefit the repeat episode produces.

The repeat episode may cost less than the first if screening and preparation can be shortened. It may cost almost as much if the service has to repeat medical checks, preparation, an all-day session, integration, and follow-up. A person returning after several years may reasonably need more reassessment than someone continuing an adaptive initial course after four weeks.

The calculator deliberately shows expected cost rather than pretending every person returns. If 40% of people receive a €10,000 repeat episode, the model adds €4,000 per person on average. A payer budgeting for 100 people must still find the staff and rooms for the forty people who physically return. Expected value smooths the budget; it does not smooth the clinic diary.

Cost-effectiveness also depends on what the return buys. If the second course restores another year of meaningful benefit, the new cost arrives with new QALYs and perhaps avoided care. If it produces a smaller or shorter response, the cost per QALY will worsen. If only previous responders are retreated and they are especially likely to respond again, the pathway may perform better than applying the same average response rate to everyone. We do not yet have robust evidence for that assumption.

This is why retreatment should not be represented only as an extra cost in a sensitivity analysis. It is another clinical transition. The model needs to show who moves into it, what care they receive, how they fare afterwards, and whether they may return again.

Once retreatment becomes possible, somebody has to decide who qualifies. A drug developer can describe treatment as episodic, but a payer still needs an authorisation rule and a clinic still needs a booking rule.

The payer might require evidence that the first course worked, a minimum period before another treatment, and a symptom score showing relapse. It might limit the number of episodes, require another trial of conventional care, or pay for the medicine while refusing to reimburse the full preparation and integration package again. Each choice changes the pathway even when the clinical protocol stays the same.

The provider faces a capacity problem. New patients and returning patients compete for the same trained staff and rooms. A clinic can appear capable of treating 500 new patients in its first year, then discover that a growing share of next year's capacity is already needed by previous patients. This is a stock-and-flow problem: every treated cohort can create some future demand.

Patients carry a different risk. People who can pay privately, travel, or remain connected to a specialist clinic may find another episode. Others may lose access precisely when symptoms return. If retreatment is clinically useful but not included in coverage, a supposedly durable treatment can produce a new inequality between people who can return and people who cannot.

The policy should therefore be visible before treatment begins. Patients should know whether further care is possible, what would make them eligible, who will monitor them, and what happens if they worsen without qualifying. A treatment pathway is not complete when it explains how people enter but remains silent about how they come back.

The unanswered questions are measurable. Trials and early services can report them without waiting for decades of routine use.

First, they can preserve the sequence for every participant: initial response, remission, later symptom change, relapse, additional care, retreatment, and outcome after retreatment. A twelve-month endpoint is much more informative when we can see the path taken to reach it.

Second, they can report the denominator at every step. How many people received the first course? How many responded? How many later met the retreatment criteria? How many were offered another episode, accepted it, completed it, and responded again? Without those numbers, a successful case after retreatment cannot tell us the probability of success.

Third, they can specify what was repeated. Did the participant receive the same dose? Was screening repeated? How much preparation and integration was added? Did the same therapists return? These details determine both the clinical experience and the cost.

Fourth, studies can collect safety and burden over cumulative exposure. A second episode may be well tolerated even when indefinite repeat treatment is not yet understood. Patient preference should be reported alongside symptoms. The fact that someone qualifies to return does not mean they want to repeat the experience.

Finally, health-economic data need to sit inside the clinical study rather than being reconstructed afterwards. Quality of life, staff time, room use, other health care, work absence, travel, and informal care can show whether the second episode produces value rather than merely another bill.

Those observations would move retreatment from a convenient model assumption to a testable part of care. They would also make it possible to compare different strategies: a fixed maintenance schedule, symptom-triggered return, booster psychotherapy, or another treatment entirely.