Michael B. First, M.D.
Columbia University Department of Psychiatry
The diagnostic categories in the DSM and the ICD are defined in terms of syndromes, i.e., symptoms that cluster together and co-vary over time. When these were first introduced into DSM-III in 1980, it was widely assumed that although the identified psychiatric syndromes consisted entirely of descriptive symptoms, their underlying neurobiological mechanisms and pathophysiology would eventually be elucidated and that one day psychiatric disorders would be defined using objective laboratory findings, as is done in most of the rest of medicine. Unfortunately, however, after 30 years of intensive efforts by the research community, not a single biological marker or gene has been discovered that is useful in making a psychiatric diagnosis . Although this frustrating lack of progress stems mostly from the fact that the problem of trying to understand the underlying etiology and pathophysiology of mental disorders has turned out to be much more complex than originally anticipated, it is likely that the categorical descriptive DSM system itself is at least partly to blame. Scientists attempting to discover the neurobiological or genetic underpinnings of psychiatric illnesses have all too often treated the man-made psychiatric constructs in the DSM as if they were “natural kinds,” looking for the gene for schizophrenia or the neurocircuitry underlying major depression as if they were real disease entities [67–69]. Perhaps whatever specificity there is between biological findings and behavioral correlates is being obscured by employing the DSM categories as if they were phenotypes, rather than focusing on more fundamental behavioral elements which cut across the various extant DSM categories.
The NIMH-sponsored Research Domain Criteria (RDoC) project is intended to establish “a framework for creating research classifications that reflect functional dimensions stemming from translational research on genes, circuits, and behavior.” (, p. 989). The RDoC project is a direct consequence of one of the aims of the NIMH 2008 strategic plan, namely, to “develop, for research purposes, new ways of classifying mental disorders based on dimensions of observable behavior and neurobiological measures.”. Using the DSM or the ICD categories as the basis for selecting research subjects invites researchers to seek a one-to-one relationship between putative mechanisms and clinically-defined disorder categories. The goal of RDoC is to instead shift researchers towards a focus on dysregulated neurobiological systems as the organizing principle for selecting study populations. The initial stage of the RDoC project is to specify those basic dimensions of psychological functioning and their implementing brain circuits that have been the focus of neuroscience research over the past several decades. Since the ultimate goal of the RDoC project is to link dysfunctions in neurocircuitry with clinically relevant psychiatric conditions, a priority in the selection of domains is that they can be related to problem behaviors that can be found in the symptom lists of conventional disorder categories [72, 73]. The preliminary RDoC working draft has identified five major domains of functioning, each containing multiple, more specific constructs: the Negative Valence Systems domain which includes constructs for fear, distress, and aggression, the Positive Valence Systems domain which includes reward seeking and learning and habit formation constructs, the Cognitive Systems domain which includes constructs for attention, perception, working memory/executive function, long term memory and cognitive control, the Systems for Social Processes domain including separation fear, facial expression regulation, behavioral inhibition, and emotional regulation constructs, and the Arousal/Regulatory Systems domain which include systems involved in sleep and wakefulness.
It is important to understand that the RDoC project is not intended to function as a diagnostic classification system in the way that the DSM and ICD do. Unlike the DSM, ICD, and other medical classifications which are designed to exhaustively describe and delineate the different ways that psychiatric patients might present symptomatically in terms of conceptually high-level concepts such as disease or disorder, the RDoC project is primarily a research framework to assist researchers in relating the fundamental domains of behavioral functioning to their underlying neurobiological components. As such, for each of the constructs noted above, the current state-of-the-art measurements/elements at several different units of analysis will be listed, including genes, molecules, cells, circuits, behavior and self –report . Thus, in concrete terms, the RDoC framework is being implemented as a matrix, with the constructs forming the rows and the various units of analysis forming the columns. In order to fill in the various cells in the matrix, NIMH is in the process of convening a series of conferences involving experts from each of the domain areas for the purpose of refining the list of domains and constructs; this includes providing working definitions, as well as compiling for each unit of analysis a listing of the measures and components that contemporary research has identified as pertaining to a particular construct.
The RDoC approach represents a true paradigm shift in the classification of mental disorders, moving away from defining disorders based on descriptive phenomenology and instead focusing on disruptions in neural circuitry as the fundamental classificatory principle. Whether RDoC ultimately bears fruit in terms of eventually improving clinicians’ ability to predict prognosis or treatment response will depend on how well this new approach performs for research , something that will takes years or even decades to fully realize. But every long journey begins with a first step; taking a fresh start in the way the RDoC project is planning to do is clearly the way to go.
Ronald Pies, M.D.
SUNY Upstate Medical University Department of Psychiatry
Even for those of us with plentiful ego supplies, outlining an alternative to the DSM system in fewer than a thousand words poses quite a challenge! Accordingly, I would refer the reader to my more detailed comments in Bulletin 2. In that essay, I proposed the following: 1. Changing the name of our classification scheme to the Manual of Brain-Mediated Disease, or MBMD; 2. Conceptualizing MBMD conditions as “instantiations of disease” in so far as they entail substantial suffering and incapacity (dis-ease) – not as reified entities in a physical sense; 3. Separating clinical descriptions of disease (“prototypes”) from research-oriented criteria; and 4. Drawing upon six foundational principles (“The 6 Ps”), as follows:
"“Privilege” refers to strict limitations on what sort of conditions are permitted into the diagnostic schema: i.e., only conditions that entail
substantial intrinsic suffering and incapacity
would be “admitted” into the MBMD."
"“Prototypes” refers to the use of
or archetypes of disease, rather than
“dimensional” methods of classification."
"“Pragmatism” refers to the
nature of the diagnostic schema; specifically, psychiatric diagnosis is seen fundamentally as a
toward the effective relief of certain kinds of human suffering and incapacity."
"“Parsimony” refers to the goal usually expressed in terms of Occam’s Razor; i.e., “entities should not be multiplied beyond what is necessary.""
"“Pluralism” refers to the use of multiple types of evidence and levels of understanding in deciding what ought to count as instantiations of brain-mediated (“psychiatric”) disease."
Finally, “Phenomenology” –i.e., the contents and structure of the patient’s felt experience—would be an important part of the prototypical descriptions in the MBMD.
Now, I am keenly aware that the designation “brain-mediated disease” creates philosophical problems for many scholars, and this term is not to be confused with the designation brain disease. (Disease, as I have argued elsewhere, is properly predicated of persons, not of tissues or organs. See [35, 74, 75]). My colleague, Michael A. Schwartz MD, has suggested the alternative term “psychiatric disorders” (personal communication, Jan. 17, 2011), and others might prefer the title, “Manual of Neuropsychiatric Disease.” In my view, either term would be a vast improvement over the troublesome Cartesian vestige, “mental disorders.”
How might my proposal work for a particular condition, such as schizophrenia? Here is a greatly truncated representation of the MBMD prototype for schizophrenia:
Sal is a 30-year-old single male whose chief complaint is “I can’t find pieces of me and the pieces I do have are fading, fading, fading, inter-dimensionally.” Sal’s problems began when he was about 14. According to his parents, Sal began to withdraw from friends and schoolmates and “seemed to enter a world of his own.” He became increasingly unable to maintain his hygiene, school performance, or social relations, often spending days at a time secluded in his room and refusing to shower. He would eat only foods that had been “de-contaminated from radiation,” which he believed was being “beamed” into the house. By age 18, Sal complained of “voices eating away at my brain,” and described hearing several persons discussing him in derogatory terms while alone in his room. At times, Sal would laugh or giggle inappropriately, as when attending the funeral of a family member…
Of course, the “real” MBMD would provide a much more detailed and comprehensive prototype. When warranted by epidemiological and clinical data, prototypical descriptions of disease subtypes would be provided; e.g., “paranoid” or “catatonic” subtypes of schizophrenia (though arguably, these subtypes are not adequately supported by existing studies). In addition, under the rubric of “Ancillary Findings,” the MBMD would provide data on prototypical neuropsychological findings; brain imaging and related neurophysiological data; and a rich description of the patient’s “inner world,” such as provided by Silvano Arieti in his classic text, Interpretation of Schizophrenia. The general nature of the “suffering and incapacity” would be described in detail for a given condition. There would be no specific “necessary and sufficient” criteria for applying a diagnosis, in the manner of the “3 from column A, 2 from column B” approach of the DSMs; rather, the clinician would be prompted to decide if the overall description of the prototype and its ancillary findings generally fit well with the patient’s history, mental status exam, and experience of the world. A list of exclusion criteria would also be provided; e.g., “the patient’s condition is not more likely due to brain injury, intoxication, substance abuse,” etc., though these could well be co-morbid diagnoses. In short, the prototypes would be “fuzzier” than the research diagnostic criteria provided in a separate part of the MBMD, but would be consistent with those criteria. (Readers old enough to remember the DSM-II (1968) will probably be amused at the superficial resemblance of my MBMD prototypes to descriptions in the DSM-II – ironic, but quite intentional! However, unlike the DSM-II descriptions, the MBMD prototypes would be written from a patient-based perspective).
Finally, I would like to see the MBMD simplify the overall psychiatric disease classification. I believe that the vast majority of clinically-significant instantiations of brain-mediated disease could fit into one of the following general categories:
Disturbances of Mood, Affect and Emotion
Disturbances of Reality Perception and Psychic Integration
Disturbances of Attention, Cognition and Memory
Disturbances of Appetitive Behavior or Impulse Control
Disturbances of Interpersonal Relations and Social Adaptation
Disturbances with “mixed” or overlapping features
Ultimately, I would envision an integration of these general categories with emerging data on endophenotypes, biological markers, and brain neurocircuits. But this should not diminish the emphasis the MBMD will give to phenomenology: the unique “felt experience” of the patient.
I use the term “brain-mediated” in two senses. The “loose”, ordinary-language sense holds that a condition or disease is brain-mediated if most salient aspects of the condition are explained by, or associated with, the function and dysfunction of the brain, and not some other organ. On this view, both schizophrenia and epilepsy are instantiations of “brain-mediated disease.” Alternatively, the “strong form” of the argument holds that a condition is “brain-mediated” when the brain is both necessary and sufficient to account for the relevant “inputs” and “outputs” of the condition. “Inputs” refer to the proximate etiological factors (e.g., abnormal dopamine levels, head trauma, etc.) leading to disease; “outputs” refer to the experiential and behavioral expressions of the condition (e.g., hallucinations, delusions, seizures, etc.). The “strong” definition – for reasons beyond the scope of this piece – is clearly harder to defend than the ordinary-language version.
Thanks to Michael A. Schwartz MD for his reading of an earlier version of this piece; and to Robert Daly MD, for his stimulating essays on the nature of “madness”.
Further readings: [35, 74, 75, 77–82]
Joel Paris, M.D.
McGill University Department of Psychiatry
The introduction of widespread dimensional measures into DSM-5 involves this as well as the previous question. Revisions of the DSM system must give a strong priority to clinical utility. When diagnoses become too complex, they will not be made, and patients suffer the consequences. The more complicated features proposed for DSM-5, such as dimensional scales, should be reserved for research purposes. Clinical diagnosis, which is carried out by busy practitioners, benefits from streamlining and simplicity.
Specifically, the dimensionalization of diagnosis proposed for personality disorders, as well as the proposals for scales to measure functioning, and for scales that assess disorders within a wider pathological spectrum, all lack clinical utility. Clinicians will not be interested in dimensions unless they are measured in a practical way, as for example, blood pressure. Asking them to score patients on multiple dimensional scales is impractical, and as First points out , the consequences of failure of usage due to poor utility could be even more serious than leaving the system unrevised.
This problem raises another question: why does the DSM system not consider separating the use of its manual for clinical and research purposes. Clinicians do not have time to make diagnoses using complex procedures, and should not be expected to spend more than a few minutes making a decision. Researchers, on the other hand, have the luxury of using complex measure to be precise, and need to have measures that have strong scientific validation. Failure to separate these uses of the manual is a deficit that affects the entire system.
Allen Frances responds: Alternate universes must prove themselves
Every month or so, someone (usually very smart and passionate) sends me a detailed proposal for a new diagnostic system offered as an alternative to the jumbled, pedestrian, atheoretical, and purely descriptive method used in DSM. The new system is invariably theory driven, clever, neat, and plausible. Surely, it is quite easy to be more coherent than a DSM that consists of a jumble of disorders gathered together largely through a historical accreting process based mostly on clinical observation and descriptive research – without an underlying theory or deep knowledge of causality.
The new systems come in 3 types: 1) Brain biology – these used to be based on correlates with neurotransmitters, but recently neural networks of various kinds are much more popular; 2) Psychological dimensions – hundreds of scales have been developed and carefully tested; and 3) Systems based on psychodynamic, ethological, and developmental models – less popular now than they once were.
Unfortunately, none of these approaches, however elegant, is remotely ready for inclusion in the official system of psychiatric nomenclature. DSM must by its very nature be a conservative document that follows and never leads the field. The problem with all of the suggestions to replace the admitted DSM jumble is that there are so many contenders, none of which has been proven or has attained wide acceptance from the field. It is also not possible to choose one from among so many plausible, but necessarily parochial systems, when most clinicians have absolutely no interest in any of them and the proponents of rival systems can make about equally valid claims for their respective pet methods.
The DSM-IV experience with the personality disorders was a rude and disheartening awakening. I very much hoped to include a (at least optional) dimensional personality rating scale. We were able to gather together in one room the proponents of all the competing dimensional systems to attempt the selection of one or some compromise among them. It didn't work – we could not forge a consensus because each participant remained wedded to his own scale (however minimally different it was from its near neighbors). Without wide agreement, it is impossible to force a field to accept changes that represent one necessarily narrowly defined perspective. The DSM-5 effort to include personality dimensions will also undoubtedly fail – for this reason as well as for its unbelievably byzantine complexity.
I feel sure that our clumsy descriptive classification may not be the only, or even the optimal way, to sort things for future research. But I feel equally certain that the DSM remains necessary to carry forth the current, everyday, practical clinical and administrative work that are its first priority. Once we have attained a widely accepted, etiological understanding of at least some forms of psychopathology, the new insights will gradually replace our clumsy, but nonetheless now still useful system.
At this stage in this arena, the wisdom of the philosopher Vico trumps the much greater and better known Descartes. Descartes sought to use what we now call Cartesian rationality and mathematical order to sort what were previously seemingly disorderly phenomena. This turned out to be a screaming success in the mathematical, physical and chemical worlds, but has (as Vico predicted) much less purchase in understanding the sloppy complicatedness of human affairs – including psychiatric diagnosis.
Response to Dr First: I agree that NIMH's RDoC project is the best hope for revolutionary advances in psychiatric diagnosis and in our understanding of psychopathology. But the obstacles are huge. The complexity of the brain has dwarfed the reach of even our most powerful research tools. Our science will advance, but probably will uncover vast new territories of our ignorance for every new beachhead of new knowledge. It may take decades of concerted effort for this project to bear clinical fruit and impact on the diagnostic system. It is an open question whether NIMH will be able to mount the necessary sustained commitment. Many things could derail the future momentum of this project – budget constraints, a new director with different goals, a lack of cooperation among scientists, a lack of findings or findings that caste things in a different light. RDoC is indeed our most promising seed – let us hope it grows and thrives. But the prospects for its future success are unpredictable in these early days.
Response to Dr Pies: Dr Pies offers a plausible possible alternative to DSM diagnosis, but certainly not one that is compelling enough to cry out for acceptance. There are obvious problems: 1) Dr Pies' prototypal system of clinical diagnosis is a throwback to DSM II and would likely create the familiar problems of low reliability and a complete disconnect between clinical practice and research; 2) We agree that DSM may be too much a splitter's system, but Dr Pies' big lumps would fail to capture the complexity and heterogeneity of clinical presentations; and, most important; 3) there are no compelling data to support that the system would work. In the absence of scientific advance supporting and requiring change, this exercise in reorganizing our tired descriptive psychiatry feels like furniture rearrangement – not likely to further our field.
Response to Dr Paris: I agree completely re the foolishness of the DSM-5 dimensions, but place more value on having one system used by both clinicians and researchers.