From the Editor
Her housing is unstable; major relationships have ended; she is deeply in debt. She presented to the emergency department hoping for help with her crystal methamphetamine addiction. “That drug just grabs you and holds you.” No medications have demonstrated efficacy for stimulant use disorder. But could contingency management be part of a meaningful plan for her recovery?
In the first selection, a paper published last month in The American Journal of Psychiatry, Lara N. Coughlin (of the University of Michigan) and her co-authors attempt to answer that question. They did a retrospective cohort study, comparing those who received contingency management with those who didn’t, looking at outcomes and 12 months of data, and involving 1 481 patients and an equal number of people in the control group. “This study provides the first evidence that contingency management use in real-world health care settings is associated with reduced risk of mortality among patients with stimulant use disorder.” We consider the paper and its implications.
In the second selection, Tony Rousmaniere (of Sentio University) and his co-authors examine large language models as health providers. In a timely paper for The Lancet Psychiatry, they weigh the regulatory and legal contexts. “LLMs have entered everyday use for mental health. Developers who embrace transparency and collaborative research can transform the mental health landscape and define the future of digital care for the better.”
And in the third selection, Emily A. Kumpf (of Johns Hopkins University) writes personally about her first-episode psychosis in Psychiatric Services. While she is grateful for the care she received in the emergency room, she was traumatized by the experience. “When I was restrained, every part of me genuinely believed the medications they were injecting into me were chemicals intended to kill me. My scream pierced through the hospital walls; I thought I was dying. To my surprise, I woke up the next morning.”
DG
Selection 1: “Contingency Management for Stimulant Use Disorder and Association With Mortality: A Cohort Study”
Lara N. Coughlin, Devin C. Tomlinson, Lan Zhang, et al.
The American Journal of Psychiatry, 10 September 2025 Online First
Although opioid overdose deaths have begun to show modest reductions, stimulant overdose deaths continue to increase, with more than a fivefold increase in the past decade… While there have been many efforts to reduce opioid use in recent years, stimulant use has not been adequately addressed. In addition to overdose risk, due to both stimulant toxicity and also because stimulants are a vector of exposure to opioids, people with stimulant use disorder (StUD) are at increased risk for substantial psychosocial harms, such as psychosis and depression, and medical conditions, such as cardiovascular conditions and infections (e.g., HIV, hepatitis C)…
To counter these enormous impacts, widespread access to effective treatment for StUD is critical. Clear and consistent evidence supports contingency management (CM) as the most effective treatment for StUD. Because of this, and considering the surging opioid and stimulant overdose crisis in the United States, the U.S. Department of Health and Human Services recommends CM as a first-line treatment. However, CM implementation has been slow, especially in comparison with treatments for opioid use disorder.
So begins a paper by Coughlin et al.
Here’s what they did:
- They conducted a retrospective cohort study of patients with StUD who received or did not receive CM using data from July 2018 to December 2020.
- They drew on electronic health records and death records in “the largest integrated health system in the United States, the Veterans Health Administration (VHA).” All-cause mortality data were obtained from the National Death Index.
- The primary outcome: mortality in the year following the index CM visit.
- They did different statistical analyses, including calculating adjusted hazard ratios using stratified Cox proportional hazards models.
Here’s what they found:
- “A total of 1,481 patients with StUD who received CM were included alongside 1,481 matched control subjects.”
- Mortality. “In the CM cohort, 27 patients died in the 12 months following CM initiation, while 46 patients died in the control cohort (unadjusted hazard ratio=0.58…) resulting in crude incidence rates per 100,000 person-days of 4.8 deaths in the CM group and 8.2 in the control group… After adjusting for covariates, including time-varying hospitalization, the reduction in mortality risk associated with CM was 41% (adjusted hazard ratio=0.59…).”
- Hospitalizations. “In the multivariable model, CM was associated with higher risk of psychiatric hospitalization (subhazard ratio=1.48…).”
A few thoughts:
1. This is a strong study published in a major journal on an important topic.
2. The main finding in a sentence: “People who received CM had about a 40% lower hazard of all-cause mortality in the year following CM initiation, after adjusting for all included covariates, compared with the matched control cohort of those who had StUD but did not receive CM.”
3. Wow.
4. How to put that in perspective? “The magnitude of the decrease in risk for death among those who received CM compared with those with StUD in the control group is clinically important and is similar in magnitude to the reduction in all-cause mortality from treating opioid use disorder with buprenorphine…” Wow, again.
5. Let’s temper our enthusiasm, though, by appreciating the limitations of this study. As the authors note: “it was an observational study that may be subject to selection bias, although we documented differences in baseline characteristics between the cohorts and adjusted for these differences using multivariable analyses. Additional work, including prospective studies to further clarify causal relationships, is needed.” So – this paper should spark interest and further study, but not necessarily change practice just yet.
6. How should we think about contingency management? A past Reading looked at a thoughtful review on this topic. The JAMA Psychiatry paper found that CM management cost less than diabetes care. You can find it here:
The full American Journal of Psychiatry paper can be found here:
https://psychiatryonline.org/doi/10.1176/appi.ajp.20250053
Selection 2: “Large language models as mental health providers”
Tony Rousmaniere, Simon B. Goldberg, and John Torous
The Lancet Psychiatry, 9 September 2025 Online First
General purpose chatbots are routinely used for personal health questions, including mental health. Converging evidence from independent surveys, observational data from public forums, and media reports show extensive use of large language models (LLMs) by people with anxiety, depression, relationship problems, and in crisis situations. These data sources have methodological limitations, but they suggest LLM chatbots have already progressed from personal coaching into psychotherapeutic intervention. This progression raises the question of where we draw the regulatory line between a so-called online assistant and an online psychotherapist; the answer will have far-reaching implications.
So begins a paper by et al.
“Although LLMs show promise for education, triage, and support, their clinical effectiveness and safety remain insufficiently established and evaluation methods are heterogeneous. Dangerous interactions of LLM chatbots with users have already been documented. A 2025 study compared the performance of three leading LLMs with expert suicidologists using a standardised suicide intervention inventory; the LLMs showed an upward bias in judging responses as appropriate and performance varied by model. The study underscores that although models can at times approximate the performance of trained humans on narrow benchmarks, they remain inconsistent.”
“The legal landscape is shifting rapidly, and it is unclear how long the doublespeak of offering therapy but denying responsibility can last.” They add: “Product liability cases against chatbots are progressing through US courts, meanwhile the US Congress has introduced bills to restrict legal protection for algorithmic amplification and generative artificial intelligence.”
They offer several suggestions.
- Collaboration. “Collaboration between developers, clinicians, researchers, and policymakers, although not simple, will be essential to address challenges as the field moves from wellness towards health-care tools. LLM developers need such collaborations to improve their clinical performance; meta-analyses show existing digital mental health apps achieve standardised mean differences of only 0.20-0.30, whereas face-to-face psychotherapy averages approximately 0.80.”
- Trials. “Clinical trials conducted by independent researchers are needed to rigorously test the safety and efficacy of LLM-based mental-health interventions. This will require the companies that own LLMs to share de-identified interaction and outcomes data with qualified researchers. Such data are also required for observational studies to enable independent real-world characterization and evaluation of how these tools are used.”
- Guidelines. “Because these tools are already being used, practical guidelines for mental-health use of LLMs should be drafted in parallel by developers, clinicians, and researchers, with frequent revision as evidence becomes available.”
A few thoughts:
1. This is an excellent paper on a relevant topic.
2. The paper opens on a cautionary note but recognizes the extraordinary moment. “The scale and granularity of LLM-interaction data offer an unprecedented opportunity to understand help-seeking behaviour, treatment effectiveness, and symptom trajectories.”
3. They make several recommendations. Is the call for practice guidelines the most important?
4. AI has been considered in past Readings. In a recent selected paper, Dr. Allen Frances (of Duke University) focused on AI and psychotherapy, arguing: “Artificial intelligence is an existential threat to our profession. Already a very tough competitor, it will become ever more imposing with increasing technical power, rapidly expanding clinical experience and widespread public familiarity.” You can find it here:
The full Lancet Psychiatry paper can be found here:
https://www.thelancet.com/journals/lanpsy/article/PIIS2215-0366(25)00269-X/fulltext
Selection 3: “The Duality of My Night in the Emergency Room”
Emily A. Kumpf
Psychiatric Services, 23 September 2025
As I walked through the doors of the emergency room (ER), my whole body shook as the voices grew louder. ‘This will help you, and the doctors will not hurt you, I promise,’ my mom told me, her voice an opposing force against my paranoia and hallucinations, which were warning that the doctors in the hospital wanted to kill me. It was August 2020, during the COVID-19 pandemic, and no one was allowed to enter the ER with me. I had to trust my mom’s voice and walk up to the front desk: just me; my hallucinations, paranoia, and delusions; and the bright, piercing fluorescent waiting room lights. My family was right: services in the ER did help, and they arguably saved my life that night. At the same time, that night also left me with intense trauma symptoms for years to follow. I have often reflected on the existing duality of my experience from the hospital that night: an immense gratitude for being able to access emergency care during psychosis and the reality of living with intense posttraumatic stress symptoms from involuntary chemical and physical restraint.
So begins a paper by Kumpf.
“There are a lot of misconceptions around psychosis – while I was in a psychotic state that evening, I was agitated and not an easy patient, by any means. However, I was never violent. This experience has made me wonder: is there a better way to de-escalate young adults who voluntarily go to the ER in extreme distress with first-episode psychosis so that restraints can be avoided altogether?”
She notes her personal connection to mental health. “At the time of this episode, I had just published my first book about suicide prevention and mental health on college campuses. I was also working full time as a counselor in a psychiatric hospital, where I had completed Crisis Prevention Institute training.”
She describes the memories. “At times, my body and mind would flash back to when health care providers were surrounding me and holding me down. In those moments, my body did not understand that the providers were trying to help. These memories would make me feel like I was trapped in time. When the flashbacks came, my body would begin to go numb, and suddenly I would feel fear wrap tightly around my limbs like a cord. My vision would start to blur, and I would need to sit on the floor to ground myself. How is it that the clinicians who saved my life from the burning sun of psychosis also appear years later, in threatening flashbacks?”
She discusses the path to the ED and her mania. “There are days and moments when I think about the path that led me to the ER, which I now recognize as mania. I remember twinges and glimmers into the preceding months of manufactured beauty. During those months, I felt my brightest. I felt like I was glowing. The colors around me seemed brighter. The sounds of music flowed through every part of my body, and each song lyric spoke to me as I paced outside for hours with my headphones on. Not a moment went by in those months when I was physically still.”
She talks about her recovery. “A large part of my recovery has been learning how to exist in the space between the sun and the ocean without getting too close to either side of mania or depression. It has been almost 5 years since that night in the ER. Since that time, I have rebuilt an incredibly meaningful life: I graduated with my master’s and now have the privilege of working on projects that focus on injury and violence prevention and community engagement in Baltimore; I have crossed finish lines of marathons, and I am active through running, kickboxing, and dance; I have deep and meaningful relationships; I am beginning to pursue a Ph.D.; and I feel genuinely like myself. I experience the ups and downs that come with life and being human without getting too close to the sun or the bottom of the ocean.”
And part of her recovery is accepting her experiences. “It took me years to shed the backpack I carried, which was filled with guilt and shame surrounding my psychotic break. When I was 23 after my episode, I searched for hours and hours online, looking for studies, similar experiences – anything to validate the duality I grappled with around my care, the feelings of shame around being restrained, and the guilt around my psychosis. Not until I asked a question of a Depression and Bipolar Support Alliance online peer support group, which I had been attending for months, about how to let go of the shame did I finally begin to rid myself of it. ‘It’s hard, but you are going to have to learn to forgive yourself for what happened during the psychosis. It wasn’t you and who you are at your core – it was your illness’ was the guidance I received. As I near the end of my 20s, I have learned the power of forgiveness, acceptance, self-compassion, and common humanity.”
A few thoughts:
1. This is a beautiful essay.
2. The description of her restraint is haunting. “To my surprise, I woke up the next morning.”
3. A vital part of her recovery: she grants herself forgiveness, a process she describes with care and insight. How many of our patients struggle with guilt about their illness? How often do we ask about it?
The full Psychiatric Services paper can be found here:
https://psychiatryonline.org/doi/10.1176/appi.ps.20250354
Reading of the Week. Every week I pick articles and papers from the world of Psychiatry.
Recent Comments