From the Editor

What has been the most significant innovation in mental healthcare delivery in recent years? It wasn’t a new medication or therapy, but the widespread adoption of the webcam in 2020. Over the course of a handful of pandemic weeks, psychiatrists and therapists switched to virtual sessions, making it easier for people to receive care, including psychotherapy, unbound by geography, and thus addressing inequity – or, at least, that was the hope. As noted recently in The New York Times: “In the 1990s, teletherapy was championed as a way to reach disadvantaged patients living in remote locations where there were few psychiatrists. A decade later, it was presented as a more accessible alternative to face-to-face sessions, one that could radically lower barriers to care.”

So, are more people receiving psychotherapy? And has this new era of virtual care resulted in better access for all? Dr. Mark Olfson (of Columbia University) and his co-authors attempt to answer these questions in a new paper for JAMA Psychiatry. Drawing on the data of more than 90 000 Americans, they analyzed trends in outpatient psychotherapy in the US, finding more care than ever before. That said, they note greater inequity: “psychotherapy use increased significantly faster among several socioeconomically advantaged groups and that inequalities were evident in teletherapy access.” We consider the study and its implications.

As doctors, we often shy away from discussing our health, especially our mental health – even with our own physicians. This is particularly concerning because doctors have a higher suicide rate than the general population, yet fears of vulnerability, judgment, and stigma keep many of us silent. In this episode of Quick Takes, I sit down with Dr. Joss Reimer, president of the Canadian Medical Association, who openly shares her own experiences with depression, as a doctor and as a patient. “We all need help sometimes.”

And in the third selection, Matthew Flathers (of Harvard University) et al. analyze AI depictions of psychiatric diagnoses in a new paper for BMJ Mental Health. They tested two AI image models with different diagnoses and commented on the results. “Generative AI models acquire biases at every stage of their development – from societal prejudice in online training data, to the optimisation metrics and safety guidelines each developer puts in place. These layered biases persist even when their precise origins remain elusive.”

DG

Selection 1: “Trends in Outpatient Psychotherapy Among Adults in the US”

Mark Olfson, Chandler McClellan, Samuel H. Zuvekas, et al.

JAMA Psychiatry, 4 December 2024  Online First

Psychotherapy is one of the most common modalities for delivering mental health care in the US. We recently reported that the percentage of US adults receiving psychotherapy increased from 6.5% in 2018 to 8.5% in 2021. However, the extent to which this overall increase varies across sociodemographic groups or levels of psychological distress remains unknown… 

During the pandemic, there was a rapid expansion of telemental health care. For many people, mental health care delivered via the internet offers a convenient, flexible, and less stigmatizing alternative to in-person care. Nevertheless, concern exists that several groups, including those with more serious mental health problems, older adults, people with lower incomes, and some racial and ethnic minoritized groups have not benefited proportionally from the recent expansion of telemental health care. Some patient barriers to telemental health care include technological challenges, preferences for in-person care, low digital literacy, financial constraints, and a lack of broadband access. Because of the far-reaching psychological and social impacts of the COVID-19 pandemic, together with the emergence of telemental health, there is increased concern over the possibility that recent trends in outpatient psychotherapy may have perpetuated or even deepened historical disparities in access to care.

So begins a paper by Olfson et al.

Here’s what they did:

  • They conducted a “a repeated cross-sectional study of psychotherapy use among adults…”
  • They looked at data from 2018 to 2021, drawing from the Medical Expenditure Panel Surveys, which are “nationally representative surveys of the civilian noninstitutionalized population.”
  • Psychotherapy was broadly defined – “a treatment technique for certain forms of mental disorders relying principally on talk/conversation between the mental health professional and the patient” and included individual, family, and group therapies. 
  • They collected data on whether the psychotherapy was in person, by video, or over the phone.
  • Demographic information was also collected. 
  • Primary outcome: differences in the use of psychotherapy and video-based psychotherapy (teletherapy) between 2018 and 2021.

Here’s what they found:

  • A total of 89 619 individuals participated.
  • Demographics. Most were female (51.5%) about half were between 35 and 64 years; the majority were White (62.2%).
  • Rise in psychotherapy. Between 2018 and 2021, psychotherapy use increased significantly faster for females than males, younger than older adults, college graduates than those without a high school diploma, privately insured than publicly insured, adults at 2 to 4 times the poverty level, employed persons than unemployed persons aged 65 years and younger and urban to than rural residents.
  • Rise in teletherapy. Teletherapy use was significantly higher among younger than middle-aged or older adults, females than males, not married than married persons, college educated adults than those without a high school diploma, people with higher than lower incomes, privately than publicly insured persons, and urban than rural residents.

A few thoughts:

1. This is a good study providing relevant and timely data, published in an excellent journal.

2. The main findings in a sentence: psychotherapy increased for the young, the more educated, those with higher incomes, and people living in urban areas.

3. The New York Times summarizes the reality in a punchy headline: “Online Therapy Boom Has Mainly Benefited Privileged Groups…” Ouch.

4. The number of adults receiving psychotherapy rose to 8.5% in 2021, up from 6.5% in 2018. (Perspective, another study by Dr. Olfson found that the annual percentage of adults taking psychiatric medications didn’t change over that time period, stable at 17.5%.) But, yes, equity didn’t increase.

5. The authors comment on the hope and the reality. “Prior to the COVID-19 pandemic, telemental health services were often viewed as a means of increasing the supply of accessible mental health professionals in rural areas, which have been persistently underserved. However, despite the rapid expansion of telemental health care during the pandemic, teletherapy does not appear to have addressed this public health challenge.”

6. Where to go from here? They make several recommendations, including: “Greater mental health literacy among more highly educated adults may help account for the observed association between education and use of psychotherapy, which has been previously reported. Mental health literacy, which involves increased knowledge concerning the characteristics of mental health problems and their responsivity to treatment, has been associated with help-seeking after the onset of mental health problems.”

7. But the problem is more than stigma. In a new New England Journal of Psychiatry Catalyst paper, Dr. John Torous (of Harvard University) and his co-authors draw on US and international data, showing that digital literacy is also a problem, perhaps the biggest problem.

That paper can be found here:

https://catalyst.nejm.org/doi/abs/10.1056/CAT.25.0009

8. Technology offers an opportunity, not a panacea.

9. Like all studies, there are limitations. The authors note several, including: “survey participants may have underreported psychotherapy use due to recall problems, stigma, or other factors”

The full JAMA Psych paper can be found here:

https://jamanetwork.com/journals/jamapsychiatry/article-abstract/2827464

Selection 2: “Living with depression”

Joss Reimer

Quick Takes, 22 January 2025

In this episode of Quick Takes, I speak with Dr. Reimer. Yes, she speaks about her approach to talking with patients and her hopes for the healthcare system. Dr. Reimer also discusses her journey: she was diagnosed with depression during training. Our conversation covers everything from her decision to speak out to her own recovery.

Here, I highlight several comments:

On her advocacy

“The stat we hear most commonly is that 20% of Canadians have a mental illness at any given time. And, you know, up to half of us will be affected by a mental illness at some point in our life, and that means that all of us either will experience it or have someone close to us who experiences it. And within the health care field, and particularly with physicians, it’s a little worse than the general population; we see higher rates of suicide amongst physicians than in the general population, for example.

“It’s something that as physicians, we don’t like to talk about. We don’t want to be vulnerable. We’re supposed to be the helpers… 

“I want to be part of a solution. I want to be part of moving us into a direction where we can value our own mental health, because when we’re not well, we can’t take care of other people to our best ability either.”

On physician fears

“I know a lot of people are scared about what their Colleges are going to say if they find out that they have a mental illness. My experience with my College is that they’ve been very supportive, that they want physicians to do well and to thrive. They don’t want to punish you for seeking help, but just make sure you have the supports that you need. That doesn’t mean that it’s not a scary thing to think that the group that’s regulating you might regulate you.

“Like I said, physicians have higher suicide rates than the general population. And that’s far worse than having to deal with your College.”

On thriving

“I have depression as a diagnosis. I’ve had it for a very long time. It’s part of who I am. It’s a chronic illness. It’s something that I have to manage every single day. And I want people to know that you can have a mental illness and still thrive with the right treatment, the right management, the right supports.”

On her depression

“The imagery that was always in my head: I was in a hole in the ground and looking up at the world. Everybody’s up there and all of the tasks I need to do (just simple things like grocery shopping), I have to first climb out of a hole. And that’s how I felt – like everything was too big and too far away and not reachable.”

On her experience

“My depression – as hard as it’s been for me many times in my life – has also led me down a path to a career that I love and I find energizing and inspiring.”

The above answers have been edited for length.

The podcast can be found here, and is just over 25 minutes long:

https://www.camh.ca/en/professionals/podcasts/quick-takes/qt-jan-2025—dr-joss-reimer-cma-president-on-living-with-depression

Selection 3: “AI depictions of psychiatric diagnoses: a preliminary study of generative image outputs in Midjourney V.6 and DALL-E 3”

Matthew Flathers, Griffin Smith, Ellen Wagner, et al.

BMJ Mental Health, 4 December 2024

The advent of generative artificial intelligence (AI) image models is rapidly reshaping our digital landscape with profound implications for the creation, dissemination and interpretation of mental health information. This transformation is significant given the well-documented impact of visual portrayals on public perceptions of mental disorders, help-seeking behaviour and stigma production.

As generative AI tools become increasingly accessible, patients are more likely to turn to these tools for clinical information. This is particularly true in mental health fields where barriers to care often drive individuals to seek alternative sources of support. It is essential for clinicians to be cognisant of these developments. In the coming years, clinicians will likely encounter patients who have been influenced by AI-generated depictions of mental illness and a greater appreciation of how AI image models behave will enable them to proactively engage in conversations with these patients.

Research on the specific impact of AI-generated images in mental health contexts remains limited. A 2024 scoping review of generative AI in mental health identified over 30 relevant studies, yet none reported on AI-generated images.

So begins a paper by Flathers et al.

Here’s what they did:

  • The authors investigated “how state-of-the-art generative artificial intelligence (AI) image models represent common psychiatric diagnoses.”
  • They prompted two generative AI image models, Midjourney V.6 and DALL-E 3 
  • They used “diagnostic terms for common mental health conditions.” They selected these terms “to highlight a diverse array of common mental health conditions listed in the WHO’s fact sheet on mental disorders.”
  • “The resulting images were compiled and presented as examples of current AI behaviour when interpreting psychiatric terminology.”

Here’s what they found:

  • “The AI models generated image outputs for most psychiatric diagnosis prompts.” (They add: “While we observed recurring themes, no formal consistency analysis was conducted.”) See below for some of the schizophrenia-generated images. 
  • Culture. “Generative AI models reflect cultural perceptions of mental disorders rather than evidence-based clinical ones.”
  • Biases. “AI image outputs resurface historical biases and visual archetypes.”

A few thoughts:

1. This is a good and unique paper, with an unusual but important look at AI.

2. The main finding in a sentence: “[Generative AI images] reflect current perceptions, historical inaccuracies and linguistic biases – a combination that can shape public understanding of psychiatric conditions.” Ouch.

3. Dr. Niall Boyce (of the Wellcome Trust) comments in his substack: “The results are not pretty, and in the case of ‘narcissistic personality disorder’ look like they are more informed by the Batman franchise than the DSM.”

4. How to address the situation? The authors suggest that “generative AI companies hire clinicians to expand on existing prompt manipulation techniques to minimise harm.” They argue: “Today, a user’s text prompt is often augmented by generative image tools to make the output more aesthetically pleasing, invisibly appending phrases to the input the model itself receives. Clinicians could help generative AI companies translate user-input prompts into safer backend prompts that have been tested and proven to produce less biased and stigmatising output results.”

The full BMJ Ment Health paper can be found here:

https://mentalhealth.bmj.com/content/27/1/e301298

Reading of the Week. Every week I pick articles and papers from the world of Psychiatry.