We've updated our Privacy Policy to make it clearer how we use your personal data. We use cookies to provide you with a better experience. You can read our Cookie Policy here.


The Death of Diagnosis: Can Research Find a Better Way To Understand Mental Health?

The Death of Diagnosis: Can Research Find a Better Way To Understand Mental Health? content piece image
Listen with
Register for free to listen to this article
Thank you. Listen to this article using the player above.

Want to listen to this article for FREE?

Complete the form below to unlock access to ALL audio articles.

Read time: 12 minutes

In the aftermath of World War II, scientific revolutions reshaped our planet. Rocket technology provided a blueprint for our exploration of space. The Manhattan project rapidly accelerated our ability to use particles – too tiny to comprehend – to unleash almost unlimited power and destruction. The first computers began to hint at a future where the human mind would no longer be the most powerful tool at our fingertips.

Amid these well-documented paradigm shifts, changes were happening within psychiatry that would alter medicine forever.

A mental health revolution

These changes were tied to and influenced by the wanton destruction experienced across the globe. World War II had provided clear evidence that poor mental health does not discriminate; anyone, regardless of status, could experience symptoms due to the events around them. This idea wasn’t new, but the huge numbers of military personnel, estimated at nearly 10% of the 11 million US combatants in WWII, that later became hospitalized for neuropsychiatric problems, required the field to adapt in response and move its focus beyond caring for only the severely mentally ill.

Part of the response to their seismic changes was Medical 203. This document was the brainchild of Brigadier General William C. Menninger, part of an influential family of psychiatrists that included his brother Karl. Medical 203 came to heavily influence two documents that would guide psychiatric diagnostics for well over half a century – the first edition of  the Diagnostic and Statistical Manual of Mental Disorders (DSM) and the sixth edition of the International Classification of Diseases (ICD) index – the first to be administered by the World Health Organization and the first to classify nonfatal diseases.

While these documents would drastically change over the next 50 years – psychoanalytical theory stripped out and ignorant classifications of homosexuality as a mental illness excised – their influence and size only expanded. By 2000, with the release of a revised version of the DSM-IV, called the DSM-IV-TR, there were 365 different diagnoses packed in, a figure that was nearly halved when the most recent version, the DSM-V, was unveiled in 2013.

By the time of the DSM-V’s release, the text had, as the primary handbook used by psychiatrists and clinicians to diagnose mental health conditions, become a complicated and controversial endeavor. The American Psychiatric Association, which Menninger became president of in 1948, administered revisions to the DSM, a process that involved over 160 different clinicians by the time of the DSM-5. The final product produced much debate within psychiatry, with concerns about reliability and over-medicalization of normal behavior proving incendiary. One leading critic was the influential American psychiatrist Allen Frances. Frances excoriated the DSM-5 before and after its publication, writing of his concern that “many millions” of people with normal thought patterns could be labeled as “psychiatrically sick”. Frances was, incidentally, the chair of the process that produced the previous version of the DSM. Within psychiatry, pleasing everyone has become an impossible task. But amid this debate, a new proposal was emerging, one that aimed to be as revolutionary as Menninger’s Medical 203 nearly 70 years prior.

“Mental health research was sort of stagnating”

While the DSM-5 was being developed to aid clinicians, the real areas in need of support at the time, says former National Institute of Mental Health (NIMH) acting director Bruce Cuthbert, were the laboratories exploring the origins of psychiatric disorders – a field of research called psychopathology. “Mental health research was sort of stagnating,” he says. At the time, research proposals relied on the same diagnostic manuals as clinicians did. While these tools had proved to be satisfactory – if far from perfect – for organizing complex patient groups, they were at odds with modern research practices.

Thomas Insel, the NIMH’s then-director, saw the diagnostic categories as out of step with emerging findings from clinical genetics and neuroscience research. The borders of diagnostic categories, intentionally left fuzzy to allow for clinical judgement, were producing superficially similar groups of patients that responded in totally different ways to treatments. While antidepressant compounds reliably benefit about 25% of patients that wouldn’t have gotten better using a placebo, a third of patients show treatment-resistant depression that does not respond to pharmacological intervention.

Insel envisioned a future in which studies, rather than including all patients diagnosed with a specific disorder, would stratify their samples using pre-selected brain circuits and particular genetic signatures.

Cuthbert points to cancer research as a model example. Where the field once grouped cancers by simply the organs affected,  sub-classes of cancer now monitor their response to immune therapies and specific gene mutations. In some western countries, this more selective approach has seen cancer survival double over the last 40 years. Could a similar approach to mental health research produce improvements in our mental health treatments?


The result of Insel’s plans was a framework called the Research Domain Criteria (RDoC), which was visualized in a matrix (Figure 1). The RDoC’s complicated structure reflects the complexity of the system being studied. It breaks down the effects of psychiatric illness on the brain into six domains – including cognitive systems, positive valence – response to beneficial stimuli, like rewards – and negative valence – reactions to situations that induce fear or sadness. RDoC then tasks researchers with exploring these systems at different levels of biology, stretching from genetics and molecules up to behavior and self-report assessments, encouraging researchers to join psychological constructs to the brain circuits that underlie those behaviors. “An example is reward-related activity, which is linked to the ventral striatum,” says Cuthbert. “We can start tying together all these different ways of looking at mental health.” Further, RDoC asks researchers to incorporate individuals with mild or subclinical symptoms into their study, placing mental illnesses at the end of a spectrum that extends through to “normal functioning”.

Figure 1: An adapted look at the RDoC matrix, which links particular domains of brain function to different levels of scientific interrogation. 

A decade of progress?

The RDoC was officially unveiled by the NIMH in 2013. After nearly 10 years, says Paul Holtzheimer, a psychiatrist and professor of psychiatry and surgery at the Geisel School of Medicine at Dartmouth College, there remains a “fundamental tension” in the field.

Holtzheimer’s research sits at a cutting-edge intersection of neuroscience and psychopathology. One focus is on electrical neuromodulation, where low-level currents and magnetic fields are used to change how neurons fire. It’s a fascination that Holtzheimer has held since the late 1990s, when he worked alongside Helen Mayberg, now a professor of neurology at the Icahn School of Medicine. Mayberg had earlier identified, using developing neuroimaging technologies, how previously hidden areas of the brain acted in people with treatment-resistant depression. “She had started work showing that this specific brain region, the subgenual cingulate, seemed to be involved in treatment-resistant depression specifically,” Holtzheimer explains. Mayberg showed that the implantation of electrodes deep into the brain could produce clinical benefits in some patients with treatment-resistant depression.

Twenty years later, Holtzheimer continues to use an ever-growing array of neurostimulation techniques, such as transcranial magnetic stimulation (TMS), where a wand bristling with magnetic fields is held over brain regions and the neural response monitored, to explore psychiatric illness. As the deputy director for research at the National Center for PTSD, Holtzheimer has focused on how war changes the brains of veterans, the same niche that motivated the formation of Medical 203 decades prior.

To Holtzheimer, study design in his field prior to the implementation of the RDoC meant important questions went unanswered. To explore depression, scientists would have to choose one of the several clinical definitions of the condition and find volunteers that hit the criteria. “There was the recognition,” says Holtzheimer, “That there were certain symptoms, like anxiety, that appear in a lot of different diagnoses. But there were very few studies that truly looked at some of these symptoms transdiagnostically.”

“Patients with schizophrenia present with depressive symptoms, as do patients with post-traumatic stress disorder (PTSD) and obsessive-compulsive disorder. I think depression may be something that represents a final common pathway of whatever is going on inside the brain,” Holtzheimer explains. By limiting trials of “depression” to individuals who have a diagnosis of major depressive disorder and nothing else, he argues that psychiatry might limit the scope of research.

The RDoC, emphasizes Holtzheimer, has freed researchers from this diagnostic tunnel vision. It has enabled them to look at whether people experiencing, for example, anhedonia (a profound loss of pleasure), have the same brain circuity underlying that specific symptom – regardless of whether they have been diagnosed with depression, panic disorder, PTSD or if they are undiagnosed. The RDoC approach, Holtzheimer says, could also benefit many of the veterans he sees who have been labeled with several different psychiatric diagnoses at once.

“When you really talk to the patient, you see that they probably just have one thing wrong with their brain that is expressing itself in these different ways,” he explains. “We give them all these diagnoses, say borderline personality disorder, but yet fundamentally, you could take a step back and say, well, it’s really just a sort of emotional dysregulation.”

A divided field

Even if the RDoC can help researchers, the question of whether researchers want the framework’s assistance remains unanswered. At the NIMH, Cuthbert heads up a small unit that advises researchers taking an RDoC approach to their study design. This implementation is far from mandatory, however, focusing on the division’s translational research efforts. “Around 40% of our grants [in that area] involve RDoC in one way or another,” explains Cuthbert. It’s perhaps surprising that the majority of research at the NIMH remains locked on the diagnostic systems that the RDoC had tried to shift. To Holtzheimer, this is indicative of a stubborn mindset within some granting bodies. “What gets complicated is that study sections at NIMH - the groups that review the grants and determine priority of studies - still think very much diagnostically,” he explains.

Cuthbert admits that there are “different opinions” on how research is conducted within the NIMH but reiterates that the institution’s policy remains not to turn down any grants, even those still purely based on DSM principles.

If there are divides within the NIMH on how effective RDoC is, the “tension” that Holtzheimer mentions is apparent in the wider field as well. “I think there’s been a lot of pushback on RDoC,” says Cassie Boness, a research assistant professor specializing in clinical psychology at the University of New Mexico. “It’s not all that helpful in terms of giving us concrete steps about how to fulfill its strategic goals.”

Flaws in the framework?

Some of that pushback has been more overt. A commentary published by Christopher Ross and Russell Margolis, both professors of psychiatry and behavioral sciences at Johns Hopkins University, levels several criticisms at the RDoC. While the approach has “value for understanding normal human psychology,” write Ross and Margolis, “for the most serious of mental illnesses… we argue that RDoC is conceptually flawed.”

The duo’s main criticism is that the RDoC's spectrum-based approach to mental health is overlooking fundamental changes in the brain that are unique to certain disorders. They also characterize the RDoC as a top-down approach to research that, in focusing on the connection between brain functions, subsystems and circuits, has become disconnected from medical models. Cuthbert rejects this view of the system. “We’re asking ‘What have the basic sciences told us about mechanisms that we can then build on for translational research?’. To us, that’s bottom up,” he suggests. 

These two viewpoints are indicative of psychopathology’s place at the nexus of multiple competing disciplines. The brain’s seemingly infinite complexity means that the things that go wrong with it can be studied from computational, molecular, epidemiological, behavioral and many other perspectives. If some psychiatrists have rejected the RDoC’s neuroscience-first approach to organizing research efforts, others, like Boness, have tried to navigate a new path that integrates the RDoC’s principles.

A third way for psychopathology

Boness’s research revolves around addiction, an area where our understanding of the connection between neuroanatomy and behavior has leaped forward in recent years. Boness, however, has remained frustrated by the DSM-5’s definition of conditions like alcohol use disorder, which she sees as imprecise and uninterested in exploring any underlying biology.

Considering the DSM’s origins, that’s hardly a surprise – they are clinical tools, not research manuals. “With the DSM, they really wanted to come up with a method by which these symptoms could be easily assessed across different types of providers in a way that was reliable and observable. You can get self-report information from somebody about these consequences or these symptoms. We’re taking a step back and saying, whoa, these consequences or these “symptoms” could be determined by other things in someone’s life, not just their alcohol or substance use,” says Boness.

From these observations, Boness has developed the Etiologic, Theory-Based, Ontogenetic Hierarchical (ETOH) Framework. ETOH is based on mechanistic understandings and includes domains, subdomains and components within them. Unsurprisingly, Boness and her co-authors explicitly reference the RDoC in their design. But she says ETOH is able to look more closely at the unique challenges within addiction.

Figure 2: An adaption of Boness et al's ETOH Framework for addiction. Credit: Boness et al, adapted from https://osf.io/bscuh/

Boness is not the first to try and refine RDoC in this fashion – a neuroclinical approach called the Addictions Neuroclinical Assessment (ANA) has already been developed by researchers at the National Institute on Alcohol Abuse and Alcoholism. But Boness believes that her approach can improve on this previous initiative: “[The ANA] is really premised on one predominant theory about addiction and how it develops. I want to take a step back and more broadly look at the literature to try and approach this more systematically without any allegiance to one theory,” she says.

This more specific approach yields several interesting insights. “We identify negative emotionality as really key,” explains Boness. In some of the literature we are seeing, dysfunction in that domain really predicts how well somebody does in terms of their recovery.”

The death of diagnosis?

RDoC’s true use, then, might be as a set of guiding principles that direct researchers to think about our mental health in new and specific ways that have previously been overlooked. That process might happen, in the end, without using the RDoC matrix itself at all.

Nevertheless, for some fields, Ross and Margolis’s skepticism may ultimately be proved correct. Holtzheimer, despite backing RDoC for his own study of depression, thinks that some psychiatric illnesses may prove beyond the framework’s ability to link: “I think for some of the more complex symptoms, it’s a little less clear. Take hallucinations. Somebody with bipolar disorder who is hallucinating in the context of a manic episode and a patient with schizophrenia who is hallucinating constantly – maybe it’s the same biologically across those groups and maybe it’s very different. We don’t have a good way of settling that argument,” he says.  The studies that might be required to prove the RDoC’s worth to these complicated questions are likely to be huge, expensive and, in the face of hostility from grant panels, not very likely to be commissioned, Holtzheimer concludes.

“The DSM and the ICD are so established, and the disorders are so reified, that people think that they are real diseases." ~Bruce Cuthbert, former acting director of the NIMH

If RDoC, or an RDoC-like approach, can eventually unite warring factions within mental health research, what would the next steps look like? How might it lead to changes in how mental disorders are diagnosed in the clinic, not just how they are investigated in the lab?

Cuthbert, for one, is acutely aware of the task such an effort would involve: “The DSM and the ICD are so established, and the disorders are so reified, that people think that they are real diseases. It will take a long time for the field gradually to shift their ideas and think differently, and for the clinicians who are used to diagnosing just by interviews in an office. But as we continue to move along, we will see changes to the actual diagnostic manuals, the DSM and the ICD. The increasing consensus in the field that these are heterogeneous disorders that need to be understood in a different way will penetrate into the clinical manuals, which in turn will help drive clinical practice,” he says.

But Cuthbert also sees optimism that the ideals of RDoC are beginning to take hold in unexpected places. He highlights the European Union’s Innovative Medicine Initiative’s Psychiatric Ratings using Intermediate Stratified Markers (PRISM)  project as an “exemplar” of RDoC principles in practice. The project’s focus on social withdrawal led to the development of quantifiable biological parameters for this behavior that are relevant across schizophrenia, major depression and even Alzheimer’s disease.

Back in the US, the reliance on DSM diagnoses may prove harder to shift for a more cynical reason. While Holtzheimer acknowledges that, in practice, many clinicians prescribe based on symptoms – giving antidepressants to someone reporting symptoms of depression, even if they don’t meet exact criteria – he also says that, at the end of a session, patients will come away with a diagnosis. Insurers will simply not cover services rendered by a clinician unless the patient has a diagnosis. “A lot of the drive towards diagnosis in the United States and Canada is driven by a monetary need, that if you're going to get paid, there needs to be a diagnosis,” says Holtzheimer.  Even if RDoC does come to ultimately dominate research approaches into mental health, the $1.1 trillion US health insurance industry might ensure the death of diagnosis remains some time away.