Dr. Strangelove, or how we learned to stop worrying and love uncertainty

“We demand rigidly defined areas of doubt and uncertainty!” ¹

If the 1980s and 1990s are commemorated, in the short history of psychiatry, for the revolution in the taxonomy of mental illness, the last decade will be remembered for the struggles with the by‐product we have created over this reform: diagnostic silos as the organizing principle for mental health care ² .

After years of dominance of the psychoanalytical formulation, psychiatry embraced the medical model in the DSM‐III. This was a necessary paradigm shift that achieved some of its goals to a large degree (e.g., increasing reliability, improving communication among clinicians and researchers, establishing the ground for empirical research), while failing to deliver some of its promises (e.g., validity and the discovery of the [biological] origins of mental illness).

Recent years have seen growing dissatisfaction with the DSM. Researchers have criticized its atheoretical and agnostic essence, with no reference to “brain‐based” concepts or “psychological” constructs. Clinicians have complained about the lack of meaningful clinical utility for case management and treatment selection, with many observed clinical cases either falling under several diagnostic categories or not easily fitting into any. Patients and their families have objected to the “mechanic” operationalized reductionist procedure that ignores the individual. What once was a celebrated revolution has become the scapegoat – the culprit for all our failures, almost.

Do we have a real exit strategy from this Greek tragedy? The paradox is that we fail to generate new knowledge in a system that is irreplaceable without new knowledge. Given that research and clinical practice have different needs and priorities, several alternative frameworks have recently been proposed: the research‐oriented, Research Domain Criteria (RDoC); the model‐driven, network approach towards psychopathology; the all‐purpose, Hierarchical Taxonomy of Psychopathology (HiTOP); and the utilitarian, transdiagnostic clinical staging. In this brief piece, we will follow a pragmatic approach and attempt to discuss how we can at least mitigate the issue of diagnostic silos in clinical practice by applying a few adjustments until we establish a better diagnostic system, ideally a pathoetiology‐based taxonomy.

One shall tolerate uncertainty to embrace pluripotency. In contrast to the culture of science, where absolute confidence is considered a cardinal sin, the culture of medicine often fails to acknowledge uncertainty: (scientific) hypotheses vs. (practical) diagnoses ³ . Notwithstanding the deep‐rooted uncertainty, psychiatry makes no exception. Early psychopathology consists of a pluripotent mixed bag of phenotypic expressions that follow diverse trajectories defying traditional diagnoses. However, psychiatry has constructed balkanized frameworks modeled after traditional diagnostic silos, such as the clinical high‐risk concept4, 5. More recently, the transdiagnostic clinical staging model has been introduced to capture heterogeneity in clinical and functional outcomes, in order to improve prediction and prevention of illness progression ⁶ .

In comparison to the RDoC and the HiTOP, the transdiagnostic clinical staging model appears to be motivated by a pragmatic clinically‐oriented mindset, and can therefore be easily and readily integrated to the current clinical practice and further applied in quality enhancement projects to iteratively test and improve practice at youth mental health services. However, there are important caveats to this model that require some further thinking.

First, it represents yet another categorization (in the temporal domain), the boundaries of which remain to be determined, with implicit referencing to etiological distinctiveness and clinical relevance, for which no more proof exists – or a priori should exist – than for traditional categorization. Second, importantly, the structure and the semantics instantly remind us of cancer. The staging system should allow for a bidirectional (up and down) shift between stages as opposed to the currently proposed progression (unidirectional) model. In this regard, the current staging model implies that mental suffering is devoid of plasticity. This is a strong assumption that is difficult to support by current scientific evidence. In addition, the transdiagnostic progression model would not be practical, given the fact that one‐fourth of the population would ultimately end up reaching at least transdiagnostic stage 2, even though the use of the clinical staging system is being limited to the age group 12‐25 years ⁷ . Also, many would agree that we should avoid associating mental disorders with cancer, which would add further negative connotations – imagine using the staging terminology to communicate with young patients and their families. Nevertheless, the staging system may at least help to a degree with overcoming diagnostic silos in practice.

One shall decrease quantity to increase utility. No current classification system has the claim for diagnostic categories as representatives of true distinct entities, yet diagnoses have been reified over time. Furthermore, the number of mental disorder categories has increased with each new edition of the DSM – so‐called diagnostic inflation – even though accumulating evidence shows that mental disorders lack clear boundaries, with large phenotypic and pathoetiological overlap. It is questionable how often many of these categories are used in routine clinical practice. Broad umbrella spectrum disorder diagnoses, such as psychosis spectrum disorder, enriched with a transdiagnostic dimensional assessment of symptoms and functioning, may suffice ⁸ .

One shall characterize to personalize. Psychiatry should embrace its limitations and uniqueness in medicine and return to the roots by putting the “person” at the center. The advent of the DSM has devalued clinical characterization and inadvertently reduced the case formulation into a standard operating procedure, easy yet insufficient. As discussed in a recent review ⁹ , we should “pay more than lip service” to better clinical characterization, that should go beyond a mere symptom checklist. Psychiatry, like other branches of medicine, is an art form that applies science in practice. The classical art of psychiatry has not been “cool” for a long time; the focus of the “clinical” psychiatry training curriculum should, nevertheless, be on psychiatric interview skills and clinical reasoning based on the characterization, until research delivers algorithms that can support or automate parts of the clinical reasoning.

One shall collaborate to alternate. Academic psychiatry should invite a wide range of stakeholders (e.g., patients, their families, carers, mental health practitioners, and policy makers) to actively take part in this process from the beginning, by identifying key issues and proposing solutions to meet the needs of our society.

Until convincing evidence is provided, the current classification system is unlikely to be superseded by the proposed alternatives for use in clinical practice. In the meantime, the above adjustments may help to overcome the issues arising from diagnostic silos in psychiatry.