Key moments, in order
The hierarchy is badly undertaught: every time we ask psychiatry residents, even at reputable residencies, at least 80 percent have never heard of it.
DSM-III's makers were divided between prioritizing validity and prioritizing reliability. The pragmatic choice was reliability first: get people to agree on what to call things, then study validity. The rigorous validity testing planned after publication really did not happen, and the little research that was done was ignored.
Before DSM-III, American psychiatrists were diagnosing schizophrenia about twice as often as their British counterparts, and a lot of what we would now call bipolar disorder and borderline personality disorder was being called schizophrenia.
Validity is how well a diagnosis tracks, or maps onto, an actual disease or condition that exists in reality, biological or psychological.
A diagnosis is valid when multiple validators, independent lines of evidence, converge: symptoms, genetics or family aggregation, the longitudinal course of illness, and biological markers, with treatment response as a stand-in for markers. The greater the convergence, the more valid, or real, the condition.
The number of diagnoses jumped. The Feighner paper listed 14 psychiatric disorders and the Research Diagnostic Criteria of 1978 listed 25, but DSM-III, published just two years later in 1980, listed 265 distinct diagnoses.
Restore the focus on validity and put it ahead of reliability. Reliability is important, but you get it when you are studying real things.
Much of this, we believe, is spurious or false comorbidity, which a diagnostic hierarchy would help resolve. Long lists usually result in excessive polypharmacy and symptom-based descriptions, and inappropriate polypharmacy is rampant in patients with borderline personality disorder.
Systems such as HiTOP (the Hierarchical Taxonomy of Psychopathology) are built around symptoms, but symptoms are only one of the classic validators. In general medicine, many diseases share symptoms like cough, fever and pain.
One proposal rates each diagnosis on its empirical validity: the more robustly validated disorders get priority, and you move down the hierarchy only if the patient does not meet criteria for them. It puts validity ahead of reliability, because reliability is important but comes when you are studying real things.
Patients now often arrive with four, five or six psychiatric diagnoses, which before DSM-III was exceptionally rare. DSM-III and especially DSM-III-R removed a lot of the hierarchy, making it possible to diagnose generalized anxiety disorder and major depressive disorder, or autism and ADHD, in the same patient.
When a patient has symptoms from several chapters of the DSM, the most immediate step is to step back and think harder about what might be driving all of them. Psychiatrists used to apply Occam's razor and look for the single best explanation; that does not mean comorbidity does not exist, or that it is necessarily just one illness.
Long diagnosis lists drive it. Many patients now arrive with four, five or six psychiatric diagnoses, which before DSM-III was exceptionally rare, and that usually results in excessive polypharmacy and symptom-based descriptions; inappropriate polypharmacy is rampant in borderline personality disorder.
Patients with borderline personality disorder frequently get diagnosed with multiple Axis I conditions, when often it is all the result of the core borderline disorder. A long list of diagnoses usually brings symptom-based descriptions and excessive polypharmacy, and inappropriate polypharmacy is rampant in this group.
Without the hierarchy, a trainee presents a patient with bipolar disorder, MDD, ADHD and GAD who now cannot focus, with no frame of reference for how to think through the problem or which medication to start. Patients end up on higher and higher doses of stimulants, or running through antidepressant after antidepressant.
Use the hierarchy together with a differential diagnosis. The job is not to know for certain what the diagnosis is, but to have a prediction, a percentage for how well it fits, and an order.
Think of a cough: a patient who tells their primary care doctor they have one is not told they have a cough disorder. We have to figure out what is going on underneath.
Inattention is a symptom, like a cough. Having a cough does not mean you have cough disorder. Do not go straight for ADHD: rule in or out the other conditions you have evidence for.
It is maybe a hundred times more important now to have a diagnostic system that teaches while it serves as a scientific and clinical tool, because people use it to learn rather than learning to use it.
The DSM was never really meant to be a textbook of psychiatry or psychopathology. Before DSM-III no one really used it; it was really for research purposes.
DSM-IV was explicitly set up as a very conservative document, with a very high bar for making any changes.
We think it is underrated how much our lack of diagnostic reliability, and so of treatment reliability across the field, undermines us as a field. With a wrong or heterogeneous set of diagnoses, the right treatment does not come in a timely way and people do not get better.
A list like bipolar disorder, MDD, ADHD, GAD and lupus leaves clinicians with no frame of reference for how to think through the problem or which medication to start. With the wrong or a heterogeneous set of diagnoses, the right treatment is not delivered in a timely fashion and people do not get better.
There is an emphasis on transdiagnostic markers and symptoms, and a real skepticism that there are natural kinds in psychopathology, actual different diseases or conditions.
Schizophrenia and bipolar disorder share about a third of their associated genes and some symptoms, but diverge on the other two thirds and have completely different longitudinal courses. Treatment response differs too: give lithium to a patient with bipolar disorder and they get better; give it to a patient with schizophrenia and nothing really happens. Look across validators.
Lithium response is one of the validators that separates bipolar disorder from schizophrenia: give it to a bipolar patient and they get better; give it to a patient with schizophrenia and nothing really happens.
Shared symptoms or shared genes do not mean there are no distinctions in psychopathology. Two disorders can share a third of their associated genes and some symptoms, yet diverge on the other two thirds and have completely different longitudinal courses.
When you read that schizophrenia is just a cultural illness, come to the inpatient ward: with the more severe diseases you see there, there is no question that these are categorical problems.
Psychiatry may not be well suited to optimizing mental health, much as a cardiologist is not the person for becoming an elite runner.
Treating only mild illness hides the full range of mental illness, which skews views of whether disease categories are real.
Working in only one setting makes you say things that are objectively untrue. The state hospital's definition of bipolar is very different from the outpatient one, and treating the most severe inpatient picture as the only real bipolar disorder is a big mistake.
Do not hesitate when the evidence is there. Nobody says they do not like to jump to bipolar disorder or schizophrenia; go off the clinical evidence.
Go from diagnosis to treatment: identify what is going on before you move to how to help. Have a descriptive diagnosis before jumping to a psychological formulation, because lots of things cause depressed mood, and a formulation makes no sense if the patient has a brain tumor.
Make sure the diagnosis is correct. Calling polypharmacy the rule risks giving permission to over-medicate personality disorders, or to stack lithium, Depakote and Lamictal when the diagnosis really needs reassessing.
Diagnosis matters here. Particular diagnoses carry real implications for understanding transference and countertransference, and if you are not thinking this way you may miss what is going on right before your eyes.
For a therapist, the diagnosis that matters is the formulation. We wonder how many therapists in the community actively formulate and deliver that formulation to the patient within the first four to six sessions.
The hierarchy matters in therapy too: if something fundamentally more medical or biological is going on that will not respond to talk therapy alone, ignoring it means not much progress.
A bare label can obscure. Call it generalized anxiety disorder or PTSD and ignore the personality and developmental components that belong in the formulation, and the diagnosis robs you of vision rather than giving you clarity.
Make a descriptive diagnosis before jumping to a psychological formulation, because lots of things cause depressed mood: it makes no sense with a brain tumor or a melancholic depression, and a severe recurrent unipolar depression is, we think, primarily a biological disease. Psychological formulation makes a whole lot of sense for abandonment depression, the chronic dysphoria of borderline personality disorder, or what the old psychoanalysts called neurotic depression, which is probably a valid diagnosis.
When the depression is the chronic dysphoria seen in borderline personality disorder, or what has been called abandonment depression or neurotic depression (probably a valid diagnosis), psychological formulation makes a whole lot of sense.
Ranked by validity, the organic diseases such as Alzheimer's disease come first. Beyond them come schizophrenia, manic-depressive illness (not necessarily the same as bipolar disorder), severe recurrent unipolar depression (melancholia), chronic low-grade depression that seems part of the patient's personality, OCD, classic autism (a bit different from autism spectrum disorder) and addiction. There is some question whether PTSD is as valid as something like schizophrenia.
DSM-III mixed severe recurrent unipolar depression (melancholia) and neurotic depression (chronic, low-grade depression that seems to be part of the patient's personality) together into major depressive disorder, and that has caused a lot of problems, because you do not treat them the same way. A melancholic, severe recurrent unipolar depression is, we think, primarily a biological disease, and jumping into a psychological formulation makes no sense there.
Among the personality disorders, borderline and antisocial are the two that are pretty robustly validated, though personality disorders may belong on the second level rather than the top.
Late in his career, Gunderson argued for separating borderline and antisocial, the major personality syndromes, from the rest: list those on Axis I and let the trait psychologists dimensionalize everything else. Narcissism and obsessionality (OCPD) sit on dimensions, while borderline and antisocial stay distinct syndromes.
There may also be a hierarchy within each major modality of psychopathology (mood, personality, thought disorder and so on). Within the schizophrenia spectrum, schizophrenia would sit at the top, with conditions such as delusional disorder lower because they may not be as robustly validated.
Keep it usable. The more complex a hierarchy is, the more accurate it probably is, but the less people will follow it, and one the average clinician cannot understand will not get used.
Aim for a broad knowledge of psychopathology across all of humanity, so you stay aware of how your own practice, which sees only one slice of it, biases you.
Objections to ranking bipolar illness high likely relate to the overdiagnosis of bipolar disorder in children about 20 years ago.
Do not just stick to the DSM. So much has been written in this field over more than 100 years that we limit ourselves by reading only the latest edition and a favorite textbook. For the history, Hannah Decker's The Making of DSM-III, by a historian rather than a clinician, is excellent.
McHugh and Slavney's Perspectives of Psychiatry teaches methodological pluralism: use different models for different patients, with no one size fits all.
When a patient carries many diagnoses, step back and think a little harder about what might be driving all of the symptoms. That does not necessarily mean it is just one illness. One old estimate is that maybe 80 percent of patients have a single form of psychopathology and 20 percent some comorbidity; it may not be based on research, but maybe it is a good rule of thumb.
A personality disorder can sit behind several apparent diagnoses. Ask whether it is causing depression in the face of rejection or abandonment, anxiety in interpersonal situations, or trouble sleeping because relationships are up and down.
A strong family history changes how you read anxiety and insomnia: a patient whose father, grandfather and uncle had bipolar illness, who is always cycling, anxious and sometimes cannot sleep because the mind is racing, may have a subsyndromal bipolar condition.
Do not give every DSM diagnosis a patient meets criteria for. The idea is tempting, but it is a very harmful way of thinking, philosophically and scientifically flawed.
Diagnostic categories are social constructions, developed to try to capture real aspects of the real world. A patient may meet criteria for several diagnoses and really have only one disease.
Patients now often arrive saying they have disorder X, Y and Z, heard about on TikTok, and that this is exactly who they are; try to take that away and you face a lot of resistance. Go slow and listen, and once there is a good enough alliance, introduce the idea that maybe they do not have six different psychiatric illnesses.
Ideally, reformulating a patient's diagnoses happens over time, when there is a good working alliance and the patient trusts your judgment and senses you know what you are talking about, not necessarily at an initial evaluation. In medication management you have to deliver it rather quickly, because it has medication implications: the patient will ask why you are taking off their stimulant and benzodiazepine.
Ask for a free-form narrative and steer away from clinical terms, which mean different things to different people. When someone says depression, anhedonia or increased anxiety, ask what they actually experienced.
Social media clips about diagnoses are, per studies, often blatant misinformation.
When a patient leads with a label, slow down and ask for the specific problems they are having. It can take several tries to get past the label to the problems.
Do not blame patients for arriving as experts in their diagnosis; society and the practice of medicine gave them that impression. They are the experts in their lives and their experiences, and we need to respect that and draw it out of them.
Diagnoses are simple names we use to guide, understand and categorize people quickly, but the narrative experience of what people are going through is still the fundamental thing.
A half-joking rule: MDD, GAD and ADHD in the same patient is a cyclothymia. Maybe that is true 50 percent of the time.
Hyperthymia, cyclothymia and dysthymia are basically subsyndromal mood disorders found in the family trees of patients with major mood illness. They are very easily mistaken for personality disorders.
Hyperthymia can be mistaken for narcissistic personality and cyclothymia for borderline personality, and the misdiagnosis can go both ways.
Do not just stick to the DSM. So much has been written in this field over more than 100 years that we limit ourselves by reading only the latest DSM and a favorite textbook.
What we have struggled with is the lack of a good literature to point to: we have a decent flow chart of treatment options for each diagnosis, but not for the temperaments.
There is a real question whether temperaments should be treated at all, and whether they are actually pathological. It makes a lot of sense, perhaps, not to treat some patients with affective temperaments.
A mood temperament on its own probably should not be treated with a mood stabilizer or an antipsychotic, though in some patients with mood temperaments there is a risk of developing full-blown mood disorders.
If something is subsyndromal, below a research-grade level of pathology and severity, be very loath to prescribe medication: there is not enough evidence for it.
If you do prescribe for a subsyndromal picture, be more willing to stop, and focus on counseling, talk therapy referral and lifestyle change when function or distress is affected.
Each moment is a point made in the episode, linked to the second it was said. The topic links open everything the podcast has said on that topic.