Analyzing a classification system: the Diagnostic and Statistical Manual (DSM)#

I’m coming to this assignment with the perspective that the DSM classification system is flawed and sometimes harmful. Based on my experience conducting neuroscience meta-research (research on research), I think one of the biggest problems is the simple assumption that these diagnostic labels are actual things that can be researched. It’s also probably helpful to share that my views have been shaped by my personal experiences in the Canadian mental health system.

The assignment#

The goal of this assignment was to describe and analyze a classification system used in the real world. Rather than outline the minute workings of the classification system itself, the idea was to examine its cultural situation in terms of its creation and governance, and to summarize the main weaknesses and strengths of the system.

I chose to focus on the Diagnostic and Statistical Manual (DSM).

Background of the DSM#

What is it? The DSM is a diagnostic manual that is used by clinicians to give individuals diagnoses of mental disorders.

What resources/concepts does it label and organize? At its core, the DSM is a classification system of mental disorders. I want to emphasize that the things being classified by the DSM are mental disorders, not individuals. Of course, the importance of the DSM comes from the fact that it contains the labels given to people, and so I won’t ignore this fact. But with respect to individuals, the DSM is more like a thesaurus, because each individual can have multiple mental disorders (e.g., both Major Depressive Disorder and Generalized Anxiety Disorder; these would be called “co-morbid disorders”). In the DSM, it is the constellations of symptoms that distinguish mental disorders as separate classes (e.g., insomnia + low mood = depression, but insomnia + intense worry = anxiety).

Why is the system important? This manual has many real-world impacts. The classification system impacts not only the lives of individuals with mental disorders, but also their social network, the institutions that (hopefully) support them, and the economy at large (e.g., through the costs of prevention, treatment, and missed work). As I’ll discuss later, it also impacts mental health research.

The beginnings of the DSM. The first edition of the DSM was published in 1952 by the American Psychological Association. It was based on a publication that the World Health Organization had created after World War II, when more attention was paid to the categorization of psychiatric conditions in order to better capture the experiences of veterans. Since then, there have been multiple editions (DSM History, n.d.):

• 1968: DSM-II (main update was that it eliminated the term “reaction”, that is, the notion that mental disorders are reactions)
• 1980: DSM-III (main updates were explicitly outlining diagnostic criteria and trying not to explain the causes of mental disorders)
• 1987: DSM-III-R (main update was to clarify the diagnostic criteria)
• 1994: DSM-IV (main update was to use research literature to support modifications)
• 2013: DSM-V (main update was to expand the empirical backing of the DSM, including pointing out areas where more research is needed)
• 2022: DSM-V-TR (main update was to ensure that sensitive topics like racism and sex/gender were discussed with more sensitivity)

What are its mechanisms of governance and change? The DSM is published and maintained by the American Psychological Association. The current change process looks like this (Frequently Asked Questions, n.d.):

• Mental health professionals can propose changes based on evidence;
• More than 200 selected subject matter experts evaluate the evidence behind proposed changes and make their own suggestions;
• The Ethnoracial Equity and Inclusion Work Group reviews the manual to make the text more sensitive to diversity-related factors;
• The DSM Steering Committee reviews and approves proposed changes; and
• The APA Assembly and Board of Trustees gives final approval.

Analysis of the DSM#

Strengths and obvious issues. To explain what I mean when I say that mental disorder classes are not actual things, consider how the labels are given to people. On the one hand, classes overlap. For example, insomnia (trouble sleeping) is a symptom of both Major Depressive Disorder (MDD) and Generalized Anxiety Disorder (GAD). On the other hand, individual classes can contain a lot of variation. For example, one person with MDD might have insomnia and weight gain, and another person with MDD might have hypersomnia (excessive sleep) and weight loss. This is possibly more problematic when we consider that the former (insomnia and weight gain) is a constellation of symptoms with greater typicality in the diagnosis of Major Depressive Disorder. The seemingly-opposite constellation of hypersomnia and weight loss is given the same classification, even though it is very rare.

If having a DSM diagnosis means those two people with MDD can both get the support they need, then the DSM is doing its job in one sense. Such diagnoses are often required for prescriptions, tax benefits, and educational supports.

However, consider what this means if you want to research the brain basis of MDD; is it an actual thing that can be studied? We can only study people, not mental diagnoses. Things get even more challenging when we recognize that many people have multiple (“co-morbid”) mental disorders.

In neuroscience, we have not had a lot of success in linking brain features to DSM diagnoses. Many researchers have moved toward linking brain features to symptoms (e.g., insomnia) rather than diagnoses (e.g., GAD or MDD). I wanted to investigate how this shift might possibly influence the DSM’s classification scheme; for this, I thought I needed to understand how the DSM is governed and updated. But the more I think about how these diagnoses are used in larger systems (like government, insurance, and education), the more I wonder whether there can ever be drastic change in the classification system itself without drastic change in those larger systems.

How this all relates to bibliographic classification. Unlike books, or even images or artifacts, ‘constellations of symptoms’ are distinctly intangible. In thinking about this problem, it has helped me to imagine that each constellation of symptoms is a list in a (very short) book, and that the book is the ‘thing’ being classified. In the shelving area for Major Depressive Disorder, we would have one book containing the list [insomnia, weight gain, ...], and another book containing [hypersomnia, weight loss, ...].

Perhaps the fact that I need to stretch the situation so much to understand it is a sign that this is not the ideal use case for classification.

References#

DSM History. (n.d.). Retrieved December 15, 2025, from https://www.psychiatry.org:443/psychiatrists/practice/dsm/about-dsm/history-of-the-dsm

Frequently Asked Questions. (n.d.). Retrieved October 22, 2025, from https://www.psychiatry.org:443/psychiatrists/practice/dsm/frequently-asked-questions

Commentary#