Chapter 9: Classification in Psychiatry

Welcome to Last Minute Lecture.

This free chapter overview is designed to help students review and understand key concepts.

These summaries supplement not replaced the original textbook and may not be redistributed or resold.

For complete coverage, always consult the official text.

Welcome to the Deep Dive, where we take the sources you share with us and distill them into the most pertinent essential knowledge you need to be well -informed.

And today, we're tackling something pretty fundamental in medicine, maybe even a bit frustrating sometimes, how we classify the human mind.

We're working from a source that really dives deep into psychiatric classification systems.

Yeah, it charts the whole journey, right, from older lists right up to the DSM we use now and even what might be coming next.

Exactly.

And our mission here is to sort of walk you through that, paying special attention to this core challenge that practitioners face every single day.

The source material hits it right off the bat, doesn't it?

This big problem in psychiatry.

Yeah.

How do you classify complex stuff like behavior, emotion, when you often don't have those objective biological markers?

Right.

No blood test for depression, no x -ray for anxiety, usually.

So you're stuck relying on descriptions, what people report, how they behave.

Precisely.

And the chapter uses this great metaphor.

It calls these classification systems blurry lighthouses.

Yeah, guiding clinicians through the stormy sea of clinical practice.

Well, it's honest about the limitations, you know.

They're not perfect GPS coordinates.

Okay, so we're going to explore that evolution, how we went from relying more on, say, expert authority to the structured criteria we see today.

And then look at where things might be heading with these newer dimensional models, like RDLC and the IC911, trying to get beyond just checklists.

So first things first, why even classify?

What's the point?

Well, in medicine generally, classification serves two main functions.

Communication is one.

It gives everyone a common language.

So doctors can talk to each other.

Researchers can compare notes globally.

Yeah.

Makes sense.

And the second big one is prediction.

Some of the pioneers, like Goodwin and Guz, they had the saying,

diagnosis is prognosis.

Diagnosis is prognosis, meaning the label should tell you something about the likely course of the illness.

Ideally, yes, that's the goal.

But to even get there, we need really clear definitions.

Psychiatry uses terms very specifically.

Let's start with syndrome.

Okay, syndrome.

A syndrome is basically just a pattern, a cluster of symptoms that tend to show up together.

Think like cough,

fever, fatigue.

Right.

And that pattern could come from lots of different underlying causes, right?

Flu, pneumonia, whatever.

Exactly.

It's just the observable pattern.

So how is that different from a disorder?

Because psychiatry seems to prefer disorder over disease.

Why is that?

Ah, yeah, that's a really key distinction.

A disorder signals something of medical concern and abnormality, the term disease.

Well, that usually implies we know the underlying pathology, the actual cause, the etiology.

And for most psychiatric conditions.

For the vast majority, we just don't know that specific cause yet.

So disorder is a more accurate,

perhaps more humble term.

Okay.

Syndrome is the pattern.

Disorder is the recognized abnormality without a fully known cause.

Yeah.

And then diagnosis.

Diagnosis is the clinician's judgment call, the opinion that a specific disorder is present in a patient.

And that judgment can be, what, like a yes -no thing?

It can be categorical, yeah.

Either you meet the criteria or you don't.

Black and white.

Or increasingly, it can be dimensional.

Dimensional meaning.

Right.

Like on a scale.

Sort of, yeah.

A point on a continuum.

Rating symptoms as mild, moderate, severe, for example.

We'll definitely come back to that dimensional idea later.

Okay.

Now this categorical idea leads straight into two absolutely critical concepts.

Reliability versus validity.

The source stresses.

These are separate things.

Totally separate and crucial to grasp.

Reliability is all about consistency.

If two different clinicians assess the same patient, do they reach the same diagnostic conclusion?

That's inter -rater reliability.

Okay.

So different people agree.

Right.

Or if you assess the same person a week apart, do you get the same result?

That's test, retest, reliability.

And the DSM system, really starting with DSM 3, put a huge emphasis on achieving high reliability.

Making sure everyone was using the checklist the same way, essentially.

Or much, yeah.

But just because we can reliably agree that someone ticks, say, five out of nine boxes, does that mean the thing we're diagnosing is actually real, that it corresponds to something biologically distinct?

And that is the question of validity.

Does the diagnosis actually pick out a real entity defined by its underlying cause, its etiology, its pathophysiology?

And the source is pretty blunt about this.

It is.

It says validity remains elusive for a great majority of psychiatric conditions.

We can get good agreement on the symptom cluster, good reliability, but often we can be sure it maps onto a single, distinct underlying biological problem.

That's the struggle.

So let's ground this.

What is the official DSM 5 definition of the thing they're trying to classify?

What's a mental disorder according to them?

Okay.

The formal definition describes it as a syndrome, so a cluster of symptoms, marked by a clinically significant disturbance in someone's thinking, emotion regulation, or behavior.

Clinically significant disturbance, meaning it causes real problems.

Yes, exactly.

And critically, this disturbance is thought to reflect some kind of dysfunction, psychological, biological, developmental, and it's usually associated with significant distress or disability in life, social, occupational, et cetera.

So the system kind of assumes there is an underlying dysfunction for each category, even if we haven't found it yet.

That's the operating assumption, yes.

That hope, that assumption, is what's driven classification for centuries.

And, you know, interestingly, this whole debate goes way back.

By far back.

Ancient Greece.

Plato argued for neat categorical boxes, mutually exclusive definitions.

You're either A or B.

Clean cuts.

Right.

But Aristotle, he leaned more towards a dimensional view, seeing things as overlapping, maybe needing arbitrary cutoffs to separate them.

So categorical versus dimensional.

Right.

That debate's been around forever.

Absolutely.

Still shapes the field today.

Fast forward quite a bit.

You get key figures like in the 19th century, he was all about careful description.

Observing and grouping symptoms.

Yes.

The descriptive neo -Kreipelinian approach that really influenced early American psychiatry.

Then you had Carl Jaspers pushing for a richer understanding of the patient's subjective experience, the phenomenologic approach.

Two different angles.

Definitely.

But by the 1940s in the U .S., things had gotten messy.

Thesaurus calls it a confusion of tongues.

Confusion of tongues, meaning?

Just chaos.

Different major institutions, the military, the VA,

general hospitals using the standard nomenclature.

They all had different diagnostic systems, different names for maybe the same condition.

Wow.

That sounds impossible for research, or even just basic clinical communication.

It was complete chaos.

And that chaos directly spurred the creation of the very first DSM in 1952.

The first one.

Was it like the huge book we see today?

Not at all.

It was incredibly thin, put together by just seven people, and it listed only a couple dozen actual mental disorders once you took out the neurological stuff they initially included.

The real groundwork for the modern DSM came later.

Yeah.

In the 1960s with the St.

Louis group, people like Eli Robbins and Samuel Goose, they were pioneers in trying to validate diagnoses systematically.

How did they do that?

Through successive steps.

First, define the syndrome clearly based on clinical description.

Then see how patients actually do over time follow -up studies.

And finally, look at family history.

See if it runs in families.

More rigorous.

More data driven.

Exactly.

And that focus on observable criteria on validation led directly to something called the research diagnostic criteria, the RDC in the 1970s.

This was the real game changer.

It represented a major shift away from the dominant psychoanalytic theories of the time towards these descriptive,

specific operational criteria focused almost entirely on getting clinicians to agree on reliability.

So the RDC basically set the template for DSM -3.

Precisely.

DSM -3, published in 1980, was landmark.

It fully embraced that RDC approach, introduced the operational criteria, the symptom checklist.

Turning diagnosis into more of a structured process.

Yes.

And it was so successful at improving inter -rater reliability that many people at the time felt the problem of reliability had basically been solved.

Which brings us closer to the present day and the DSM -5 published in 2013.

That sounds like a huge effort.

Oh, it was monumental.

Task forces, multiple work groups, advisory committees.

And they started with this incredibly ambitious goal fueled by the whole decade of the brain excitement.

What was the big goal?

They initially hoped to create a classification system based on biology,

find genetic markers, specific neuroimaging findings,

link diagnoses directly to etiology right from the start.

Wow, that would have been revolutionary.

True validity.

It would have.

But as the source makes clear, the science just wasn't there yet.

Those objective biological markers, they didn't materialize in time for DSM -5.

So they had to pivot.

They had to pivot back.

Back to refining the descriptive approach, trying to improve reliability further, maybe restructure things.

A necessary adjustment, but definitely a disappointment for those initial hopes.

And to check that reliability, they used these extensive field trials.

Right.

Rigorous testing in real clinical settings.

And they measured agreement using a statistic called the Kappa coefficient.

Kappa.

How does that work?

What do the numbers mean for us?

Okay.

So Kappa measures agreement beyond chance.

A Kappa below 0 .2 is generally considered poor, basically useless.

0 .2 to 0 .4 is fair.

0 .4 to 0 .6 is moderate or good.

And above 0 .6 is considered good or excellent.

Got it.

So what did those field trials actually show?

Where did the DSM -5 categories land in terms of reliability?

It was a mixed bag, frankly.

Some did quite well.

Neurocognitive disorder, for instance, had excellent agreement.

Kappa around 0 .78.

Okay, that's high.

Yeah.

And things like schizophrenia, bipolar disorder, they were in the good range.

Kappa's around 0 .460, 0 .56, respectively.

Solid reliability.

Clinicians could generally agree.

Since a but coming.

Yes.

The big but was for some really common conditions.

The standout example, the one that really caused concern, was major depressive disorder.

One of the most diagnosed conditions.

What was its Kappa?

It came in at 0 .27.

0 .27.

So just barely in the fair or acceptable range.

Exactly.

Disappointingly low.

It means two trained clinicians using the same DSM -5 criteria often couldn't agree on whether a patient met the threshold for major depression.

That really undermines the utility of the category, doesn't it?

If agreement is that low.

It certainly raises serious questions.

How useful is a category if experienced professionals can't reliably apply it?

Okay.

Besides reliability, what were the other big changes in DSM -5 structurally?

Well, a major one was getting rid of the old multi -axial system.

DSM -3 and 4 had those five axes.

Axis I for clinical disorders, axis II for personality disorders, and so on.

Right.

I remember those.

Kind of complicated.

Yeah.

It was complex and maybe not as useful as hoped.

So DSM -5 eliminated it, moving towards a simpler structure, more aligned with the WHO's ICD system, non -axial documentation of diagnoses.

And they shuffled the chapters around quite a bit, too.

They did.

Big reorganizations.

For example, PTSD and acute stress disorder.

They used to be under anxiety disorders.

Now they have their own chapter, trauma and stressor -related disorders.

Makes sense thematically.

Yeah.

And similarly, OCD got pulled out and now anchors a new chapter, obsessive, compulsive, and related disorders.

And that's where they added the new diagnosis of hoarding disorder, formally recognizing it.

Hoarding disorder, right.

Now you mentioned major depression's reliability issues.

There was also a big controversy around its criteria change, wasn't there?

The bereavement exclusion.

Oh, yes.

That sparked a lot of debate.

Historically, DSM criteria excluded you from a major depressive disorder diagnosis if your symptoms started shortly after the death of a period.

It was assumed to be grief.

And DSM -5 removed that exclusion.

It did.

The argument was that bereavement is a severe stressor that can precipitate a major depressive episode and shouldn't automatically disqualify someone from diagnosis or treatment.

But the critics.

Critics, including many social scientists, argued very strongly against it.

They felt it risked pathologizing normal human processes, essentially turning profound, understandable grief into medical illness needing treatment.

A very contentious change.

I can see why.

Okay, any other specific changes worth highlighting?

Maybe substance use.

Yeah, that was simplified too.

DSM -IV had separate categories for substance abuse and substance dependence.

DSM -5 combined them into a single substance use disorder category for each substance.

Just one diagnosis now.

One diagnosis, but with a severity rating mild, moderate, or severe based on the number of symptoms present.

Kind of a dimensional element creeping in there.

Interesting.

And what about somatic symptom disorder?

Another significant shift.

Previously, for diagnoses like somatization disorder, a key feature was that the physical symptoms had to be medically unexplained.

Right, the doctors couldn't find a physical cause.

DSM -5 removed that requirement.

Now the focus for somatic symptom disorder is on the distress and excessive thoughts, feelings, or behaviors related to the somatic symptoms, regardless of whether there's a known medical explanation or not.

So even with all these changes, refinements, controversies,

the DSM system still faces some pretty fundamental criticisms.

The source highlights a few main ones.

First, just the sheer number of diagnoses.

Yeah, the proliferation.

We went from what?

265 diagnostic codes in DSM -3 to 438 in DSM -5.

That's a huge jump.

It is.

And it leads to worries about diagnostic inflation, about splitting categories into finer and finer distinctions.

Are we creating too many labels, maybe medicalizing normal variations?

The second major criticism mentioned is the problem of arbitrary thresholds.

Right.

The checklist approach often relies on hitting a specific number of symptoms, like five out of nine for depression, or meeting a duration rule, say, symptoms for at least six months.

And those numbers, they feel a bit arbitrary.

They often are.

There's usually no strong empirical basis for why it's five symptoms and not four, or six months and not five.

And these cutoffs might exclude people with genuine early stage illness who don't quite meet the threshold yet.

And the third big one, comorbidity.

Huge issue.

It's incredibly common for a single patient to meet criteria for multiple DSM diagnoses at the same time.

Major depression plus an anxiety disorder, plus maybe a substance use disorder.

Which, if the categories were truly distinct biological entities.

You wouldn't expect that much overlap, right?

High comorbidity suggests the categories might be too broad, or that they overlap significantly, or that they aren't really carving nature at its joints, so to speak.

They aren't capturing distinct underlying problems cleanly.

Okay.

So these limitations, the struggle with validity, the arbitrary thresholds, the comorbidity, they fueled interest in alternative approaches.

Let's talk about the main ones.

First, the ICD.

Right.

The International Classification of Diseases from the World Health Organization.

Currently on ICD -11, this is the official system used worldwide for mortality and morbidity statistics, including mental disorders.

Mandatory reporting uses ICD codes.

So it's the global standard.

How does it differ from DSM?

The biggest difference is probably philosophy.

ICD prioritizes clinical utility, what's most helpful for the average clinician working anywhere in the world.

As opposed to DSM's heavy focus on reliability for research.

Exactly.

So instead of DSM's very strict operational criteria checklists, ICD -11 tends to use narrative clinical descriptions and diagnostic guidelines.

It's a bit more flexible, relies more on the clinician matching the patient to a descriptive prototype.

Interesting.

Did they try to align DSM -5 and ICD -11?

Oh yeah, there was a big harmonization effort, lots of collaboration between the workgroups, but significant differences remain, reflecting those slightly different priorities, reliability versus global clinical utility.

Okay, so ICD is the global clinical standard.

Then there's this other alternative,

RDOC, which sounds quite different.

RDOC, the research domain criteria is radically different.

It came out of the U .S.

National Institute of Mental Health and IMH, and it's important to stress it's not meant to be a clinical diagnostic manual right now.

It's purely a research framework.

A traditional DSM symptom -based category is entirely.

Instead, it tries to build a classification system from the ground up, based on basic dimensions of functioning things we know are rooted in specific neural circuits.

So starting with the brain and behavior, not the symptom list.

Precisely.

It's explicitly dimensional, looking at functions across a range from normal to abnormal, and it's organized in this matrix.

A matrix.

Yeah.

The rows are basic functional constructs, like fear, reward processing, working memory, grouped into broad domains like negative valence systems, cognitive systems, social processes, et cetera.

Okay, rows are functions.

What are the columns?

The columns are different units of analysis for studying those functions.

Everything from genes and molecules up through neural circuits, physiology, behavior, and finally, self -report questionnaires.

Wow.

So you could study, say, response to threat at the genetic level, the

behavioral level.

That's the idea.

Build up understanding from basic biobehavioral dimensions, hoping that eventually that will lead to more valid ways of classifying mental distress based on underlying mechanisms, not just surface symptoms.

The contrast with DSM is stark.

RDOC starts brain behavior.

DSM starts symptoms.

Exactly.

And that leads to the controversy, or at least skepticism.

Critics point out practical problems.

RDOC might find, say, similar frontal lobe circuit issues in someone with ADHD and someone with early psychosis.

Okay.

But clinically, the treatments are completely different.

Stimulants for ADHD, antipsychotics for psychosis.

So for actual clinical practice today, the symptom -based DSM categories, despite their flaws, are still indispensable for guiding treatment decisions.

RDOC isn't ready for the clinic yet.

But is the RDOC idea the dimensional approach influencing the systems at all?

It definitely seems to be.

Look at ICD -11.

It has adopted dimensional approaches in some key areas, moving away from rigid categories.

Like where?

Personality disorders is the big one.

Instead of the old, often criticized categorical subtypes like borderline, narcissistic, et cetera, ICD -11 uses a primary diagnosis of personality disorder defined by general impairment, and then specifies the severity mild, moderate, or severe.

You can trait domain qualifiers, but the core is dimensional severity.

Huge.

And also in psychotic disorders, ICD -11 encourages using symptom qualifiers, rating the prominence of things like positive symptoms, negative symptoms, cognitive impairment, rather than just forcing people into the old subtypes like paranoid or disorganized schizophrenia, which had questionable validity.

So dimensionality is creeping in, even into the main clinical systems.

It seems to be the direction things are moving.

Yeah.

Slowly, carefully.

Okay.

So wrapping this all up, it's been quite a journey.

We started with this basic need to classify,

saw the chaos before standardized systems.

Then the rise of the DSM, focusing massively on reliability, maybe at the expense of validity sometimes.

And now this push towards more dimensional thinking seen in ICD -11, and especially in the research framework of RDOC.

Yeah, we're kind of caught in this tension, aren't we?

Between the DSM's focus on reliable checklists, the ICD's focus on global clinical usefulness, and RDOC's ambitious quest for true etiological validity based on neuroscience.

Three different poles.

So given all that complexity, that ongoing debate,

what's the final thought for someone learning about this?

What's the key takeaway about where classification might be heading?

Well, maybe it's helpful to go back to Carl Jaspers, one of those early figures we mentioned.

He always maintained that psychiatry is inherently a hybrid discipline.

It has to combine methods from natural sciences, biology, neuroscience, with methods from the social sciences and humanities understanding subjective experience, context, meaning.

Not just one or the other.

Right.

And the source suggests the next real breakthrough might actually come from embracing that middle ground again, instead of just relying on symptom counts from checklists or waiting for elusive biological markers.

Perhaps we need to get better at gathering that rich clinical phenomenology Jaspers talked about.

Meaning.

Meaning detailed narrative assessments.

Really understanding the individual's experience in their social and environmental context and integrating that with what we're learning about biological systems.

A richer, more contextualized picture than just ticking boxes.

So the skill of careful listening, observation,

and integrating different kinds of information.

That remains paramount.

Perhaps it's still the most advanced diagnostic tool we actually have.

A powerful thought to end on.

Thank you for joining us for this deep dive.