Diagnostic Uncertainty in Psychiatry

The classification of diseases emerged originally as a necessity within the evolution of therapeutic development. The foundation of psychiatric syndromal stratification arose from the medical understanding that, generally, patients with similar signs and symptoms can be grouped together for prognostic and treatment purposes¹. Drs. Kraepelin and Alzheimer were some of the first physicians to identify mental illness as having a neurobiological basis¹, although their work was largely disregarded throughout the 1800s. In 1844, what we now know as the American Psychiatric Association (APA) was formed and entitled the Association of Medical Superintendents of American Institutions for the Insane². During the mid-20th century, as psychoanalysis became the foundational theory of psychiatric care within the U.S., there was a blatant lack of unification among physicians, with each clinical institution operating with its own diagnostic mechanism. 

Largely, at the time, common practice among psychiatrists was to disregard biological sciences, and this eventually led to the segmentation of psychiatry from other specialties¹. As medications and neurotransmitter systems were discovered, psychiatric diagnostics were entirely overhauled. Publications highlighting standardized diagnoses evolved from the first ever diagnostic manual (Statistical Manual for the Use of Institutions for the Insane³), which became the antecedent to today’s Diagnostic and Statistical Manual of Mental Disorders (DSM). The original Statistical Manual included just over 20 diagnostic categories, centered around psychotic disorders, which made sense in the asylum-based⁴ treatment setting, where the majority of psychiatrists practiced. After a total of ten iterations by 1942¹, the APA published its first version of the DSM in 1952⁵. 

With each subsequent iteration of the DSM, the focus shifted. The first two versions were made specifically to gather epidemiological data¹ on each disorder. Later on, a push was made to include measurable criteria for research, with revisions to promote greater clinical utility¹. Presently, the DSM dictates much of what is considered to be reimbursable care - what will insurers be willing to cover? 

In 2013, the DSM-5 was published, which made great strides in differentiating itself from its predecessors through the removal of the multiaxial system that had existed since the third iteration. The axes had segmented the diagnosis of mental illness across five branches⁶: 1) mental health and substance use disorders; 2) personality and developmental disorders; 3) organic medical conditions; 4) psychosocial/environmental conditions; and 5) overall functioning. Dr. David Kupfer⁷, chair of the APA Task Force for the fifth edition, spoke about his desire to include the most updated clinical diagnostic criteria within the DSM. He wanted criteria to be based on established etiology and neurobiology¹; albeit to his chagrin, there was a lack of clarity surrounding biological subtyping at the time of publication. Like other psychiatrists, Dr. Kupfer had hoped for greater breakthroughs by the time 2013 rolled around. Diagnostic changes for the DSM-5 mainly came from field samples using naturalistic study and had slightly variable reliability¹, leading to some criticism. Simultaneously, the National Institute of Mental Health (NIMH) catalyzed its own dimensional diagnostic research project: the Research Domain Criteria (RDoC). RDoC was meant to emerge as a science-based framework, targeting pathophysiology, genomics, and neuroscience, to drive the future of reliable and valid diagnosis⁸. How it materialized was a slightly different story.

Dr. Kupfer had wanted the systematic strides of RDoC to drive the next iterations of the DSM⁷. RDoC, created in 2009⁹, aimed to reduce heterogeneity in research samples (while improving our understanding of the neurobiological underpinnings of diagnoses), uncover why there is such a large risk of comorbidity with mental illness in the isolationist diagnostic approach, eliminate sole reliance on expert consensus, and increase understanding of subthreshold symptoms. Scientists and clinicians wanted to understand the full spectrum of psychopathology across dimensions. RDoC’s focus is and was to cross-sectionally study illness¹⁰, utilizing components of genetics, neuronal circuits, behaviors, environmental influences, cognition, and more. The goal? An integrative, comprehensive model that moves away from the siloed diagnostic framework we have today. This was one of the biggest steps towards precision psychiatry: the medical community yearned for a system of diagnosis and treatment based on a richer understanding of the science and psychology of mental illness. And yet, the daily practice of clinicians is no different, at least in relation to RDoC, now than it was in 2009. 

The framework initiated areas within psychiatric research that, perhaps, can later on contribute to diagnostic and treatment pathways¹¹. For now, it has not yet transformed clinical psychiatry. One reason may be that current research within the RDoC space involves further segmenting and re-imagining existing diagnoses¹¹ in favor of building a new framework, and adoption of any new concept meets friction. Alongside the NIMH RDoC enterprise, several federal projects have emerged over the last few decades in pursuit of precision medicine; the BRAIN Initiative¹² and the Human Connectome Project¹³ are some of the latest efforts to move the needle in this space. Psychiatric medicine yearns to make a breakthrough with biological, symptom, and treatment-response subtyping¹⁴ through the use of brain imaging techniques and biomarkers. Yet, adoption of a new foundation always remains a steadfast challenge in the clinical world. 

Providers are wary of a new way to practice: what does precision psychiatry mean for all mental health clinicians? In reassurance, building rapport with patients, collecting a thorough diagnostic interview, and reviewing a life history of symptoms are steps that are not going anywhere. But what if we could measure a person’s symptoms through proxies and corollaries for diagnostic stability? How would this change their prognosis? Their burden of illness? Less than a century ago, cardiologists could not measure the inside structure of the heart nor its related function¹⁴. Yet, today, ultrasound and echocardiography are daily instruments used in the diagnosis and treatment of heart disease. What if we brought this level of diagnostic security to the assessment of patients? If we could have greater insights into patients’ neurobiology and neuronal circuitry, genetics, and symptomatology, maybe we could develop not only greater diagnostic tools but more precise and targeted interventions. While we are headed towards this goal slower than we had originally anticipated, the target remains the same - better data, better treatment.

Dina Veytsman is the Medical Affairs Lead at Headlamp Health. She is a board-certified psychiatric mental health nurse practitioner and staunch digital mental health tech advocate.