6 Suggestions for DSM-6

Making the next DSM look less like a house of mirrors

The American Psychiatric Association (APA) is cautiously considering starting the revision and development process for DSM-6, the next edition of the Diagnostic and Statistical Manual of Mental Disorders. It is not yet clear if DSM-6 will receive the official approval to proceed; however, four DSM subcommittees have been established, focusing on social determinants, quality of life and functioning, biomarkers, and classification structure. Preliminary plans were presented at the 2025 annual meeting.

In this post I will make a case for six proposals that I think will help improve the state of affairs.

I am assuming, perhaps pessimistically, that radical changes in the DSM schema are unrealistic. For instance, I am assuming that DSM is not going to abandon the categorical structure in favor of a dimensional one, or that DSM is not going to become a formulation manual. I am also focusing on general points that apply broadly rather than making specific proposals about specific conditions.

1. Clarify the DSM concept of “mental disorder”

Few ideas have caused as much confusion and wreaked as much havoc on public understanding of mental health problems as the use of the term “disorder” and the official definition of mental disorder in terms of “dysfunction in the psychological, biological, or developmental processes underlying mental functioning.” [Important clarification: DSM’s use of the term “dysfunction” doesn’t mean “functional impairment,” although many people use the term with that meaning. Dysfunction and functional impairment are seen as separate and complementary features. The intended meaning of “dysfunctional” here is closer to “abnormal.”]

Jonathan Shedler identified this in a blogpost, How One Word Sidelined Psychotherapy:

“What if the DSM didn’t append the word “disorder” to every entry? What if it were presented, more modestly, as a compendium of mental and emotional difficulties for which people seek help?

Maybe we could have avoided decades of debate about semantics, politics, and ideology, and simply gotten on with providing care…

I’m a consequentialist here. The consequence of this terminology is to sideline psychotherapy—self-reflection, self-understanding, meaning-making—as serious treatment. It marginalizes psychotherapy not through evidence or reason, but through connotation. It’s conquest by “linguistic fiat.””

“Dysfunction” is an essential feature of the DSM concept of mental disorder—more essential, according to the DSM definition, than even distress and impairment, although, in real clinical practice (and implicitly, even in the manual itself), distress and impairment take precedence. Given the importance the DSM definition places on “dysfunction,” one would expect some conceptual and scientific discussion regarding how conditions such as adjustment disorder, generalized anxiety disorder, prolonged grief disorder, and autism actually reflect “a dysfunction in the psychological, biological, or developmental processes.” Such a discussion would require DSM to explicitly state what dysfunction means (presently left undefined) and how its presence can be demonstrated or debated in clinical/scientific contexts and how relevant disputes can be resolved.

My own view is that the most accurate understanding of the DSM notion of “dysfunction” is that it is common-sensical or folk psychological, and this is something that was explicitly acknowledged by Kenneth Kendler, a DSM insider since DSM-III-R, in the Conversations in Critical Psychiatry Q&A I did with him.

“Kendler: The general idea of dysfunction is common-sensical – that the relevant psychobiological system is not doing what it is supposed to do. Examples might include providing your higher centers with an approximately veridical sense of the world around you, keeping levels of anxiety roughly appropriate to the real dangers being confronted, producing mood states approximately congruent to the environmental situation, etc. DSMs have traditionally seen disorders as existing within individuals and, for example, avoided providing diagnoses for dysfunctional marriages or families. So, in that sense, the underlying disturbance is seen to exist within individuals. I do not see that definition having much of anything to do with the causes. Environmental experiences like severe childhood sexual abuse can clearly cause dysfunctional mood-modulation systems as well as a high genetic vulnerability.”

Surely this warrants an explicit acknowledgment in the manual given the widespread tendency to interpret dysfunction in terms of disease processes and concerns regarding individualization of social problems?

In my post on the history of the inclusion of “prolonged grief disorder” in the DSM, I noted that what was significant to the public about the diagnosis was not the fact that some people struggled with complex, persistent bereavement and that they may benefit from psychotherapy for it, but rather that prolonged grief was now considered a “disorder.” The psychiatric community itself was curiously silent on the significance of that characterization. I wrote:

“Well, the DSM definition requires a “dysfunction” and I’m not seeing anyone clearly acknowledge what the dysfunction here is.

To be clear, I am not saying that there is no dysfunction in PGD. I think there is a dysfunction in a very folk psychological, common sensical, value-laden kinda way; it’s this notion that something (the process of grieving, in this case) is not functioning as it is supposed to, but the norms for “as it is supposed to” invoked here are basically personal and sociocultural norms. If that is true, then characterizing prolonged grief as a dysfunction amounts to little more than: the diagnostic manual now officially recognizes that some people experience persistent difficulties associated with bereavement that are intense and substantially prolonged beyond socioculturally normative expectations. Obviously, for a lot of people, including many philosophers and psychiatrists, calling something a mental dysfunction is a much stronger claim that says something about failure of natural design, or neurophysiological alterations, etc. For them, it’s not just: “Oh, you are suffering in a way that is socioculturally atypical and we have clinical interventions that can potentially help you.””

So what exactly am I suggesting DSM-6 should do? It is, I suspect, too much to hope that DSM-6 would abandon the term “disorder,” but it can take several steps to defuse the current problems.

• Many people are unaware that the full title of International Classification of Diseases (ICD) is “International Statistical Classification of Diseases and Related Health Problems.” It would be a short step for the Diagnostic and Statistical Manual of Mental Disorders to be the Diagnostic and Statistical Manual of Mental Disorders and Related Mental Health Problems.

• Either avoid the word “dysfunction” in the formal definition of mental disorder, or make it very clear what the intended meaning of dysfunction is and how it applies to conditions classified in the manual. I don’t think it is reasonable to expect DSM to settle the thorny philosophical question of how to best define “disorder.” However, we can and should expect DSM to at least clarify how DSM understands and applies that term to the conditions classified. I believe the correct understanding of DSM dysfunction is common-sensical or folk psychological, but if authors of DSM-6 disagree with me, all they have to do is define it clearly and show how it is applied in the manual.

I wrote in my recent Asterisk magazine article on the DSM,

“All it means for a person to “have,” say, generalized anxiety disorder in the DSM sense is for the anxiety to meet certain thresholds of severity and duration, and to negatively impact a person’s life. What that means for the state of one’s biology or for one’s self-understanding is both scientifically and philosophically unsettled.”

Once this is made explicit, an understanding of “disorder” in terms of disease entities or brain malfunctions can no longer be treated as the default one, and the space for pluralistic conceptualizations opens up.

2. Indicate the degree of empirical validation for all diagnoses

Someone browsing through the pages of DSM would get the impression that all conditions classified in it have a similar sort of clinical and scientific legitimacy. Aside from conditions placed in Section III, “Emerging Measures and Models,” everything else seems to be on an equal footing. No characterization or description in terms of strength of empirical evidence is offered.

“Disinhibited Social Engagement Disorder” has the same status as “Posttraumatic Stress Disorder,” “Social (Pragmatic) Communication Disorder” has the same status as “Autism Spectrum Disorder,” and “Intermittent Explosive Disorder” has the same status as “Obsessive Compulsive Disorder.”

This is, quite frankly, malarkey. And it leads to a highly distorted state of affairs, like a peculiar house of mirrors that flattens everything big and small into the same size.

I have previously discussed how DSM relies on validators and utilitators to determine if diagnostic categories capture clinical information of interest to us. Validity and utility are not equally distributed among DSM diagnoses. Schizophrenia and Disruptive Mood Dysregulation Disorder are continents apart in terms of validation.

DSM-6 needs some meaningful way to communicate that. DSM could, for instance, create a composite index of sorts that expresses the degree to which validators converge and provide clinically relevant predictive information. It would also be important to convey the reliability of diagnoses and what we know about the stability of diagnoses over time.

3. Explain why diagnostic thresholds are what they are.

The thing about diagnostic thresholds like dimensional diagnoses in general medicine, such as essential hypertension, is that they are intended to optimize some goal. The practical goal of diagnosing “essential hypertension” is to identify people who are at high risk for serious cardiovascular complications (such as heart attack and stroke) and to offer management strategies that mitigate these risks. Once the goal is clear, we can analyze the available data, and make the optimal decision about what threshold makes the most sense and offers the best compromise between trade-offs.

What exactly does the DSM threshold for Major Depressive Disorder, (five out of nine depression symptoms, 2-week duration, etc.) optimize? I would submit that the answer is: nobody really knows.

Is it any surprise that we are all rather confused and muddled about what exactly we are trying to accomplish by following this threshold? Is it any surprise then that most clinicians don’t really care as to whether someone’s generalized anxiety has persisted for “3 month” or “7 months” for the purposes of treatment? (DSM-5 threshold is 6 months.)

In some cases the cut-offs are clearly based on considerations other than empirical. In the case of prolonged grief disorder, e.g., according to Zachar et al. (2023), the 12-month threshold was used because of “evidence that clinicians did not favor the 6-month duration in the ICD” while Prigerson told the New York Times that although researchers were confident that the disorder can be identified six months after bereavement, APA decided to define the syndrome more conservatively, a year after death, to avoid a public backlash. I don’t know what the real reason was, but what I do know is that the DSM itself gives us no explanation for this threshold, even though it owes us one.

What kind of a scientific diagnostic manual presents us with semi-arbitrary symptom- and time-duration thresholds with no formal justification? How can we even improve thresholds if we don’t know what we are optimizing?!

As the story goes, when Robert Spitzer (chairman of the taskforce for DSM-III) was asked by Daniel Carlat about why the threshold of 5 criteria for major depressive disorder was selected, Spitzer replied, “… we came up with the arbitrary figure of five… Because four just seemed like not enough. And six seemed like too much.” (see the full exchange here)

There is nothing necessarily wrong with this strategy, but the problem is the lack of transparency and the illusion that five out of nine criteria as a threshold has some special clinical or scientific status when it is simply a best guess by a committee looking at the messy results of field trials.

A best guess is fine as a starting point, but it cannot be treated as something special or sacred, and we cannot carry out entire research programs where a pulled-out-of-the-ass threshold is assumed to capture differences in brain functioning.

So what we need from DSM-6 is transparency about what evidence, if any, supports the particular thresholds for the conditions included in the manual, and why that particular threshold is what it is.

4. Acknowledge gaps in the descriptive schema

A comprehensive description of mental health problems encountered in the clinic using specified DSM diagnoses is not possible. Large swathes of psychopathology are simply not named in the DSM, either because they are subthreshold or because the DSM simply has no category corresponding to the problem. This is a big reason why “unspecified” and “other specified” (formerly NOS or “not otherwise specified” in the DSM-IV) constitute a big chunk of diagnoses in the real world. I use Unspecified Mood Disorder quite frequently in my clinical practice because many clinical presentations of mood disturbances neither meet criteria for major depression nor bipolar disorder. What are they? The answer in DSM categorical terms is basically to shrug.

This becomes a problem for two reasons:

1. If the DSM is descriptively incomplete, we need to rely on something else for those gaps. A phenomenological or clinical description using some other strategy.

2. Aspects of the healthcare system, especially in the US, at times demand specified diagnoses. A person may not be approved by insurance for treatment with certain medications or psychotherapies in the absence of a specified diagnosis.

In my discussion of “overdiagnosis,” I noted:

“… the more I practice, the more I am of the view that the DSM criteria are rather conservative. DSM falls short of a comprehensive description of psychopathological presentations, and the categories are not broad enough to capture all clinically relevant forms of psychological suffering and disabilities for which diagnosis is needed. The criteria for specified diagnoses exclude a significant number of patients who experience considerable distress and impairment in their daily lives.

People fundamentally misunderstand the impetus behind increasing rates of psychiatric diagnosis—by and large, clinicians are not imposing labels or disorder judgments on reluctant patients who would rather not have a name for what they are experiencing (although it does happen). There is a tremendous demand and appetite for diagnostic labels as tools of understanding. The amount of suffering and disability is vast; our traditional folk-psychological categories have been inadequate for this task. The DSM legitimized psychological distress and disability as healthcare problems deserving of recognition and treatment. The profession was unprepared for the magnitude of clinically significant distress out there, and the diagnostic manuals have struggled to keep up.

Much clinical practice involves clinicians responding to real suffering inadequately captured by official criteria. This leads to stretching existing diagnoses or utilizing vague “unspecified” categories, not from diagnostic carelessness but due to genuine gaps in the diagnostic framework.”

What is the solution? I am of the view that the DSM should clearly note that the DSM is not a comprehensive compendium of mental and emotional difficulties for which people seek help and which clinicians treat in the clinic. This should be accompanied by a statement that the scope of clinically appropriate and clinically necessary treatment—and the scope of treatment that should be covered by public or private health insurance—is not determined by what has a specific name in the DSM.

Why are these gaps in the DSM at all? One big reason is that the DSM doesn’t take dimensionality seriously. Which leads me to the next suggestion…

5. Include HiTOP in the DSM appendix

DSM categories have certain advantages when it comes to clinical communication and matching available treatments to clinical indications, but they rest on arbitrary thresholds, produce heavy comorbidity, and lump heterogeneous presentations. Diagnoses can change as symptoms drift above or below cutoffs. The Hierarchical Taxonomy of Psychopathology (HiTOP) starts from data instead of historical categories based on clinical observation and organizes symptoms into statistically derived dimensions and spectra, giving psychometric structure to what DSM is trying to capture via syndromic categories.

HiTOP is underpinned by a large and coordinated research program showing that its dimensions generally offer better stability, a cleaner psychometric structure, and stronger links to validators than many DSM diagnoses. HiTOP lets us talk in terms of degrees across core dimensions and symptom profiles, which is a better match for the complexity and variation we see in the clinic.

The DSM appendix is a reasonable place for HiTOP, an acknowledgment that better, dimensional science exists and deserves an official foothold. That’s where the Alternative Model for Personality Disorders lives at the moment (hopefully it will be made official in DSM-6). Placing HiTOP there would legitimize training, research, and practice that are already moving in a dimensional direction. It would also provide a bridge between categorical and dimensional models, giving the field a shared translational map.

Including HiTOP in the appendix would be a way for the DSM to acknowledge that it is not the one, true classification. DSM is not a reflection of categories that exist in the Platonic realm of Forms. Rather, it is one historically contingent, fallible effort with advantages and disadvantages, and other legitimate classifications are possible.

[For a clinical introduction to HiTOP, see Ruggero et al, 2019]

6. Improve transparency and accountability via free access and public disclosure of potential conflicts of interest

A document as important as the DSM needs to be freely available online to the public, like the ICD is. Accessibility is a good reason on its own but the bigger reason is that it facilitates transparency and accountability. What is hidden behind a paywall cannot be easily scrutinized. The financial interests of the American Psychiatric Association are not a good reason to override this ethical obligation. APA can still make money by selling print copies, app access, and DSM-related textbooks.

Another point related to public accountability is that DSM-6 needs to be transparent about conflicts of interest. Conflicts of interest of task force members should be publicly disclosed (not just disclosed internally to the APA) and a formal effort should be undertaken afterwards to determine how potential conflicts of interest may have influenced decision-making during the development of the manual. 57% of DSM-IV task force members and 69% of DSM-5 task force members had financial associations with the pharmaceutical industry. Given what I know about what happens behind the scenes, I don’t believe that changes in the DSM are driven by financial conflicts of interest, but I can see how it may look that way to outside observers, and this is an easily addressable source of public mistrust. Industry relationships are ubiquitous in psychiatric research and hard to avoid in an endeavor like the DSM that depends heavily on expert input and expert consensus; it may be unreasonable to expect that the DSM-6 task force can be free of industry associations, but we can at least be as transparent about it as possible.

---

---

See also:

The “Overdiagnosis” Confusion

DSM Disorders Disappear in Statistical Clustering of Psychiatric Symptoms

Antirealism Will Not Save the DSM From Empirical Inadequacy