evidencepublished last verified

what the DSM is for

the companion page to our DSM reference. that one asks what the manual is. this one asks what it is for — what a set of criteria actually looks like when you take it apart, what the book says about its own purpose and its own limits, why 1980 is the year that explains everything, how a category in a book turns into a payment, and the question underneath all of it: is any of this improving access, or outcomes, or both, or neither.

the reference companion — the DSM: what it is, who writes it, and why it describes symptoms instead of explaining them — covers the editions, the committees, the biomarker question and the money. this page does not repeat any of it.

227
combinations of symptoms that satisfy the DSM criteria for major depressive disorder — 1,497 once the either/or symptoms are counted separately
636,120
symptom combinations that satisfy DSM-5’s criteria for PTSD, against 79,794 under DSM-IV
62
entries in the DSM-5 classification whose job is to name what the criteria sets do not fit
0
treatment guidelines in the manual — “DSM does NOT include information about treatments,” in the APA’s own words

read this first if you have a diagnosis. the section on access is the most important one on this page and it is good news: a diagnosis is how care gets paid for, how school services get triggered, how disability gets determined. none of what follows is an argument that your diagnosis is wrong, or that you should change or stop a treatment. stopping psychiatric medication abruptly can be dangerous. if you are in crisis, call or text 988 (US), 24/7, free.

1. two people can hold the same diagnosis and share almost nothing

This is the least-explained thing about psychiatric diagnosis and the most concrete, so it goes first. A DSM diagnosis is not a threshold on a single measurement, the way a blood-pressure reading is. It is a menu: a list of candidate symptoms, a required number to be drawn from it, a minimum duration, and a separate requirement that the whole thing be costing you something.

Major depressive disorder is the clearest example. DSM-5's Criterion A asks for “Five (or more) of the following symptoms… present during the same 2-week period… at least one of the symptoms is either depressed mood or loss of interest or pleasure” [8]. SAMHSA's published comparison table states the same structure as “Five or more of the following A Criteria (at least one includes A1 or A2)” [9]. Then, as a separate criterion — Criterion B, not one of the nine — “The symptoms cause clinically significant distress or impairment in social, occupational or other important areas of functioning” [8][9].

the nine candidate symptoms of major depressive disorder, and which of them are really two symptoms wearing one label [8][9][10]
 symptomcounts either way
A1depressed mood — one of the two required core symptoms—
A2loss of interest or pleasure — the other required core symptom—
A3weight or appetite changeloss or gain
A4sleep disturbanceinsomnia or hypersomnia
A5psychomotor changeagitation or retardation
A6fatigue or loss of energy—
A7worthlessness or guiltworthlessness or inappropriate guilt
A8reduced concentration or indecisiveness—
A9recurrent thoughts of death or suicidal ideation—

Now do the arithmetic, because the arithmetic is the point. Treating the nine criteria as nine yes-or-no items, there are 227 different combinations that qualify — that is the published figure, from a study whose entire purpose was to count them [12]. Treat the four either/or symptoms as the two different things they actually are — losing weight is not gaining weight, sleeping fourteen hours is not sleeping three — and the number becomes 1,497 [10][11].

227 and 1,497 are not rival estimates. They count at two different resolutions. 227 is the number of ways to pick a qualifying set of criteria. 1,497 is the number of ways to pick a qualifying set and land on one side or the other of each either/or symptom. Anyone who tells you one figure corrects the other has not read either paper.

The consequence is not theoretical. Fried and Nesse put it plainly: “two individuals who qualify for a diagnosis of MDD may not have a single symptom in common” [15]. Olbert and colleagues went and looked, across eighteen diagnostic categories and two large patient datasets, and reported that “in most categories, 2 individuals with the same diagnosis may share no symptoms in common” — and, empirically, “For both disorders in each of the datasets, pairs of individuals who shared no common symptoms were observed” [14]. This is not a historical curiosity: a 2023 methods review in BMC Psychiatry restates it as current fact [16].

And the combinations that do occur are thinly spread. In STAR*D, the largest US depression treatment trial, 3,703 patients produced 1,030 distinct symptom profiles; about 14% of the patients had a profile shared with nobody else in the study [15]. Zimmerman's clinical sample of more than 1,500 patients hit the criteria in 170 different ways, leaving 57 of the 227 possible combinations unobserved [12].

Depression is not the extreme case. When DSM-5 rewrote the criteria for PTSD, the number of qualifying symptom combinations went from 79,794 to 636,120 — the authors called it “an eightfold expansion” [13]. That is one diagnosis, one revision, and a change in what the label can contain that no reader of the label ever sees.

the part of the manual nobody quotes: the 62 escape hatches

If criteria sets were tight, people would either fit or not. They frequently do not, and the manual has an answer for that, printed right in the classification. We counted 62 entries in the APA's published DSM-5 table of contents whose entire function is to name what the criteria sets do not fit — 24 beginning “Other Specified” and 38 beginning “Unspecified” [5]. Other Specified Depressive Disorder. Unspecified Anxiety Disorder. Unspecified Mental Disorder. They are real, codable, billable diagnoses, and they exist because the categories above them have edges that people fall off.

That count is ours, not the APA's — the APA publishes no such figure. It is a line-level count of the published classification list in the linked PDF and anyone can reproduce it in a minute [5].

what this does not mean. None of this says depression is not real, or that a person meeting five of nine is not suffering. It says the label is a wide container. That matters for one specific and practical reason: when a study reports that a treatment works “for depression,” it worked for whichever slice of those 227 combinations walked into that trial — and your combination may not be that slice. It is a reason to ask which of your symptoms the treatment is expected to move, not a reason to doubt that anything is wrong.

2. the manual is more modest about itself than its readers are

The DSM's front matter says what the book is for, and it says what the book cannot do. Both halves are worth reading, because the second half is quoted far less often than the first — including by people who invoke the manual.

How these quotations were checked. The APA's DSM Library is paywalled and blocks automated retrieval, so nothing below was taken from an APA server. Every sentence was confirmed by exact-phrase search inside the scanned DSM-5 volumes and APA Publishing's own Desk Reference in the Internet Archive full-text index; control searches on altered wording return nothing, which is how we know the match is exact [6]. All of it is DSM-5 (2013). The current edition is DSM-5-TR (2022) and we could not verify its front matter against an authorised copy, so no sentence here is labelled DSM-5-TR.

what it is for

“The primary purpose of DSM-5 is to assist trained clinicians in the diagnosis of their patients' mental disorders as part of a case formulation assessment that leads to a fully informed treatment plan for each individual.”

DSM-5 (2013), Use of the Manual [6]

Note the three limiting words in that sentence: trained clinicians, case formulation, each individual. The APA's public pages say the same thing in plainer language: the manual “provides a common language for health care professionals who diagnose mental illness and establishes consistent and reliable diagnoses that can be used in research and to aid in the development of medications and other interventions,” and it is used “in a variety of settings, including clinics, private practice, schools, hospitals, courtrooms, and the insurance industry” [1].

And one thing it is emphatically not for. “DSM does NOT include information about treatments. It is a manual for assessment and diagnosis of mental disorders and does not include information or guidelines for treatment of any disorder” [1] — the capitals are the APA's, not ours. The same statement appears in the APA's DSM FAQ for the current edition [34]. Hold onto that when you reach section 6.

what the manual says you cannot do with it

the two admissions in the front matter

Two sentences in DSM-5's own introduction do more damage to the popular understanding of psychiatric diagnosis than any critic has managed.

“In the absence of clear biological markers or clinically useful measurements of severity for many mental disorders, it has not been possible to completely separate normal and pathological symptom expressions contained in diagnostic criteria.”

“The results of numerous studies of comorbidity and disease transmission in families, including twin studies and molecular genetic studies, make strong arguments for what many astute clinicians have long observed: the boundaries between many disorder ‘categories’ are more fluid over the life course than DSM-IV recognized, and many symptoms assigned to a single disorder may occur, at varying levels of severity, in many other disorders.”

DSM-5 (2013), Introduction [6]

The manual also formally retired the idea that carving the categories finer would eventually make them clean: “The historical aspiration of achieving diagnostic homogeneity by progressive subtyping within disorder categories no longer is sensible” [6]. And the definition of a mental disorder it settled on is functional rather than biological, with two explicit carve-outs: “A mental disorder is a syndrome characterized by clinically significant disturbance in an individual's cognition, emotion regulation, or behavior that reflects a dysfunction in the psychological, biological, or developmental processes underlying mental functioning… An expectable or culturally approved response to a common stressor or loss, such as the death of a loved one, is not a mental disorder. Socially deviant behavior… and conflicts that are primarily between the individual and society are not mental disorders unless the deviance or conflict results from a dysfunction in the individual” [7].

A manual that says all of that about itself is not the manual most people argue about. The gap between what the book claims and what the culture around it claims is, in our view, the real story of the DSM.

3. it was built to count people, and rebuilt in 1980 because psychiatrists could not agree

The reference page timelines the editions. This section is the causal story underneath them, and it has two acts: a century of counting, then one decade of crisis.

act one: a census category, not a clinical tool

The APA is candid about the origin. “In the United States, the initial stimulus for developing a classification of mental disorders was the need to collect statistical information. What might be considered the first official attempt to gather information about mental health in the United States was the recording of the frequency of ‘idiocy/insanity’ in the 1840 census. By the 1880 census, seven categories of mental health were distinguished: mania, melancholia, monomania, paresis, dementia, dipsomania, and epilepsy” [3]. In 1917 the APA's predecessor body worked with the Census Bureau on a plan “for gathering uniform health statistics across mental hospitals,” which the APA itself calls “primarily an administrative classification” [3]. That manual — the Statistical Manual for the Use of Institutions for the Insane — listed 21 disorders, 19 of them psychotic, and ran to ten editions by 1942 [17]. It was built for asylums, and it described almost nobody outside one.

The war changed the population. “A much broader classification system was later developed by the U.S. Army (and modified by the Veterans Administration) to better incorporate the outpatient presentations of World War II servicemen and veterans” [3] — the document known as Medical 203, directed by William Menninger, “who, during World War II, was a brigadier-general and the head of psychiatry in the Office of the Surgeon General” [18]. Houts describes it as having been “adapted to become DSM-I” [19]. We do not give a date for it, because the record does not agree on one: Houts' title says 1943, the reprint record says 1946, Shorter says October 1945, and the APA files it under “Post–World War II” [3][18][19]. Meanwhile the WHO put mental disorders into ICD-6 for the first time, and “ICD-6 was heavily influenced by the Veterans Administration classification” [3]. DSM-I, in 1952, was “a variant of the ICD-6” [3].

So the manual's ancestry is: a census tally, an asylum ledger, and an army paperwork problem. By the 1950s the United States had five competing official classifications in simultaneous use — asylums, Army, Navy, Veterans Administration, and the American Prison Association [17]. DSM-I existed to end that, and it was 130 pages long [17].

act two: the reliability crisis, which is the whole story

Here is the thing that explains everything downstream, and it is not a scandal about drug money. Before 1980, psychiatrists routinely did not agree with each other.

The US–UK Diagnostic Project showed it most vividly. In one arm, American and British psychiatrists watched the same forty minutes of videotape of the same patient. 69% of the Americans said the patient had some form of schizophrenia. 2% of the British audience did. [26] Same patient, same tape, two professions.

Then, in 1974, Robert Spitzer and Joseph Fleiss re-analysed six prior reliability studies across eighteen diagnostic categories and wrote the sentences that got DSM-III built: “There are no diagnostic categories for which reliability is uniformly high. Reliability appears to be only satisfactory for three categories: mental deficiency, organic brain syndrome (but not its subtypes), and alcoholism. The level of reliability is no better than fair for psychosis and schizophrenia and is poor for the remaining categories” and “The reliability of psychiatric diagnosis as it has been practised since at least the late 1950s is not good” [22]. Spitzer was appointed chair of the DSM-III Task Force being formed as that paper appeared, and he told an interviewer years later that the US–UK findings “inevitably led to justifying diagnostic criteria” [26].

The strongest objection to that story, printed at full strength. Kirk and Kutchins point out that the 1974 paper's damning verdicts were interpretive labels the authors invented for their own table, and that the same numbers admit an opposite reading: “had their purposes been different, Spitzer and Fleiss could have described the state of reliability as quite good. Or they could have emphasized that in every single diagnostic category in every study, psychiatric agreement was considerably better than chance” [21]. The man who wrote the indictment of the old system then chaired the committee that built the new one. That is not proof of anything improper. It is a fact worth holding while reading the founding document of modern psychiatric diagnosis.

The intellectual machinery was already waiting. In 1970 Eli Robins and Samuel Guze had proposed five phases for validating a psychiatric category — clinical description, laboratory studies, delimitation from other disorders, follow-up studies, family studies — in a process they described as “one of continuing self-rectification and increasing refinement leading to more homogeneous diagnos[tic categories]” [24]. In 1972 the St. Louis group turned that into the Feighner criteria: explicit, checkable criteria for the small number of diagnoses they thought were actually defensible — 14 of them [23][32]. (Sources differ; 14 is what two independent peer-reviewed accounts give, and one later paper says 20 [23].) In 1978 Spitzer, Endicott and Robins expanded that into the Research Diagnostic Criteria, opening with the diagnosis of the field's problem: “A crucial problem in psychiatry, affecting clinical work as well as research, is the generally low reliability of current psychiatric diagnostic procedures” [25].

DSM-III landed in 1980 and, in the APA's own summary, “introduced a number of important innovations, including explicit diagnostic criteria, a multiaxial diagnostic assessment system, and an approach that attempted to be neutral with respect to the causes of mental disorders” [3]. Note attempted — that is the APA's word, not a hedge of ours. The psychoanalytic term “neurosis” was struck and then, after an outcry, partly restored in parentheses [20]. The count went from 182 diagnoses in DSM-II to 265 in DSM-III [20].

and the part that is about money, stated carefully

The best peer-reviewed account of why DSM-III happened does not credit a scientific breakthrough. Mayes and Horwitz argue the shift “was neither a product of growing scientific knowledge nor of increasing medicalization” but of standardisation driven by four things, two of which were financial: “mounting pressure on psychiatrists from health insurers to demonstrate the effectiveness of their practices” and “the necessity of pharmaceutical companies to market their products to treat specific diseases” [27]. Kawa and Giordano describe the squeeze in the same period: “Diminished research support from the National Institute of Mental Health (NIMH), and reduced resource allocations by the federal government and insurance providers were prevalent throughout the 1970s” [20]. And there is a cleaner regulatory mechanism than any conspiracy: the 1962 Kefauver–Harris amendment to the Food, Drug, and Cosmetic Act required that prescription drugs be shown effective “for the treatment of categorical (specific) diseases” [28]. A drug regulator that will only approve treatments for named conditions creates demand for a list of named conditions. That is a structural pressure, not a payment, and we state it as one.

Two corrections to the story you have probably heard. First, the Rosenhan pseudopatient study — the one where healthy volunteers were supposedly admitted to psychiatric hospitals — is now described in the peer-reviewed history literature as “a spectacularly successful case of scientific fraud” [29], and Spitzer had called it “pseudoscience presented as science” at the time [31]. It should not be used as evidence of anything. Second — and this is the half that gets dropped — Rosenhan being fraudulent does not make pre-1980 diagnosis reliable. Spitzer's own 1974 re-analysis is the honest evidence for that, and it is independent of Rosenhan entirely [22]. The reliability data is load-bearing. The famous study is not.

What DSM-III bought, then, was agreement. What it did not buy is stated most bluntly by a DSM-5 work-group member: “as DSM-III and IV criteria resulted from a strong desire to increase reliability (agreement across evaluators), they sacrificed validity” [32]. DSM-III's own foreword conceded the point in advance: “there is no assumption that each mental disorder is a discrete entity with sharp boundaries (discontinuity) between it and other mental disorders” [28]. The manual grew from 130 pages in 1952 to 992 in 2013 [17]. What it was built to fix was disagreement, and disagreement is what it fixed.

4. a category in a book becomes a payment through one specific mechanism

The reference page covers who profits from the book itself — the APA's ownership of the DSM rights, the inventory-sales spike in the year DSM-5 shipped, the panel conflict counts and the APA's rebuttal to them. Read that there, not here. This section is about a different and much larger flow of money: the one that runs through every claim filed in American mental health care.

Start with a correction, because almost everyone gets this wrong, including people who write about it. The DSM is not the billing system. The APA says so in four words:

“There are no DSM codes. The codes appearing in DSM-5 in addition to the ICD-10-CM codes were ICD-9 codes. These codes are no longer valid since ICD-10-CM codes replaced them in the United States starting October 1, 2015. Clinicians should use ICD-10-CM codes to submit claims.”

American Psychiatric Association, DSM Frequently Asked Questions [34]

So the division of labour is this. The DSM supplies the criteria that decide which condition a person has. The ICD supplies the code that goes on the claim form. In the APA's framing they are “companion publications”: “The ICD contains the code numbers, used in DSM-5-TR and all of medicine, needed for insurance reimbursement (reimbursement is mostly germane to the U.S.) and for monitoring by national and international health agencies” [1]. CMS confirms the legal reach: “On October 1, 2015, the health care industry transitioned from ICD-9 to ICD-10 codes for diagnoses… ICD-10 applies to all parties covered by the Health Insurance Portability and Accountability Act (HIPAA), not just providers who bill Medicare” [33].

The APA has stated the consequence in its own materials for insurers, and it is worth quoting because it is the plainest description of the mechanism anyone has published: “Clinicians use DSM-5 diagnoses to communicate with their patients and with other clinicians, and to request reimbursement from insurance organizations” [4]. The same document says DSM-5 was “developed to facilitate a seamless transition into immediate use by clinicians and insurers” [4]. Insurers are named as a design constraint on the manual, by the manual's publisher, on page one.

One correction we owe that document: it gives 1 October 2014 as the ICD-10-CM adoption date. Congress delayed it; the transition actually happened on 1 October 2015, per CMS [4][33].

where the mapping breaks, and what that costs

Because the payable codes are ICD's and the criteria are the DSM's, the two lists do not line up one-to-one — and the APA publishes the mismatches. Some distinct DSM disorders share a single code: “Because the DSM-5 diagnostic codes are limited to those contained in the ICD, some disorders must share codes for recording and billing purposes. For example, hoarding disorder and obsessive-compulsive disorder share the same codes” — both ICD-10-CM F42 [4]. Others get filed under a name that is not their own: disruptive mood dysregulation disorder, created in DSM-5, has no ICD entry, so it is billed as F34.8, “other persistent mood [affective] disorders” [4]. Binge eating disorder was billed under bulimia nervosa's ICD-9 code [4].

Practically: to an insurer's database, a person with hoarding disorder and a person with OCD are the same row. The APA's own instruction is a workaround — “the DSM-5 diagnosis should always be recorded by name in the medical record in addition to listing the code” [4] — which tells you the code alone does not carry the information.

The other thing that mapping decides is disability. The APA notes that the Global Assessment of Functioning scale, which DSM-5 removed, “was used for determinations of medical necessity for treatment by many payers, and eligibility for short- and long-term disability compensation” [4]. A number printed in a manual was, for years, a gate on money paid to individual people. When the manual dropped it, the APA had to write guidance for the payers who had been relying on it.

The honest summary of the mechanism. Nothing above shows anyone being paid to write a criterion, and we do not claim it. What it shows is structural and undisputed: in the United States, care is paid for against a coded diagnosis; the code comes from ICD; the criteria that select the code come from the DSM. Everyone downstream — clinician, hospital, insurer, employer, school district — needs a name before money moves. That is the manual's largest economic function by an enormous margin, and it dwarfs the book sales.

5. a diagnosis is how care gets paid for, and that is the strongest true thing about the manual

This section goes before the sceptical one deliberately, because it is the better evidenced of the two and because burying it would misrepresent the record. The American Psychiatric Association describes the manual as serving “a variety of settings, including clinics, private practice, schools, hospitals, courtrooms, and the insurance industry” [1]. That list is not marketing. Each item is a real gate, and a diagnosis is what opens it.

The starting problem is enormous. In 2022, an estimated 59.3 million American adults had a mental illness, and only 50.6% received any mental health treatment in the previous year [45]. Two decades earlier the National Comorbidity Survey Replication found 41.1% of twelve-month cases had received any treatment, and concluded that “most people with mental disorders in the United States remain either untreated or poorly treated” [46]. Against that baseline, anything that reliably moves people from untreated to treated is doing real work.

the four gates

the precision almost everyone gets wrong

Here is where the easy version of the access argument falls down, and the true version is more interesting. None of those four systems legally requires a diagnosis. Every one is written functionally.

So why does everyone experience a diagnosis as the price of entry? Because of one sentence, and it is the most important sentence in this section. The same civil-rights guidance that says no diagnosis is required also says:

“OCR will presume, unless there is evidence to the contrary, that a student with a diagnosis of ADHD is substantially limited in one or more major life activities.”

US Department of Education, Office for Civil Rights, July 2016 [40]

That is what a diagnosis actually buys: a presumption. With it, the burden shifts to whoever wants to say no. Without it, you can still get through — you just have to build the functional case from scratch, every time, in front of every institution. A diagnosis is not the key. It is the master key. You can pick the lock without it; it is slower, it costs more, and most people give up. That framing is ours, but every clause under it is a citation.

the other side of the ledger, at full size

A diagnosis is not costless, and the harms are measured rather than asserted. A meta-analysis of 28 experimental studies found that framing a condition biologically “reduce[s] blame (Hedges g = −0.324) but induce[s] pessimism (Hedges g = 0.263)” and increases endorsement of the stereotype that people with psychological problems are dangerous (g = 0.198) — with the authors' own caveat that the last result “could reflect publication bias” [47]. Their conclusion: “Promoting biogenetic explanations to alleviate blame may induce pessimism and set the stage for self-fulfilling prophecies that could hamper recovery” [47].

And a 2026 systematic review of 26 studies of adults diagnosed with autism later in life found both things at once — its two organising themes were “Constantly Struggling” and “Revelation” — concluding: “While diagnosis can grant understanding, acceptance, and community, autistic adults continue to struggle with stigma, regret, and inadequate support post-diagnosis” [48]. Relief and cost, in the same finding, from the same people.

6. the category is reliable. that it predicts what will help you is much less clear.

The honest answer to Adam's question — is it improving outcomes? — is that we largely do not know, and where the question can be checked, the category does less work than its reputation implies. That is not the same as saying it does none. Here is the evidence in both directions, at the same size.

Start with the sitting NIMH director's own framing, from 2013: “The strength of each of the editions of DSM has been ‘reliability’ – each edition has ensured that clinicians use the same terms in the same ways. The weakness is its lack of validity” [49]. Reliability is agreement. Validity is being right about something. They are different achievements and only one of them has been demonstrated — a point our reference page works through with the DSM-5 field-trial numbers, which we do not repeat here.

evidence the categories are not carving nature at the joints

A claim we went looking for and did not find. We wanted a peer-reviewed sentence stating flatly that DSM category is a weak predictor of treatment response. It does not appear to exist. The literature makes the point obliquely — Insel's “symptoms alone rarely indicate the best choice of treatment” [49], the p-factor paper's “challenging to find causes, consequences, biomarkers, and treatments with specificity to individual mental disorders” [52], the APA's schizophrenia-subtype concession [55] — and never flatly. So we do not assert it either. This paragraph is the honest shape of an absence.

evidence the category absolutely does change treatment

This half is not a concession, it is the record, and getting it wrong is dangerous.

“are these labels stable?” is the wrong question

Stability is not a property the DSM has or lacks. It is a property of each individual category, and inside psychosis alone it runs from almost-always-right to a coin flip. The largest pooled estimate — 42 studies, 45 samples, 14,484 first-episode patients, average follow-up 4.5 years — ranks them, and the spread is the finding [78].

how often a first-episode diagnosis was still the diagnosis at follow-up, by category. 1.00 would mean it never changed [78]
first-episode diagnosisstill holds95% CI
schizophrenia0.900.85–0.95
affective spectrum psychoses0.840.79–0.89
schizoaffective disorder0.720.61–0.73
substance-induced psychotic disorder0.660.51–0.81
delusional disorder0.590.47–0.71
acute and transient / brief psychotic disorder0.560.62–0.60 (as printed)
psychosis not otherwise specified0.360.27–0.45
schizophreniform disorder0.290.22–0.38

The interval printed for acute and transient / brief psychotic disorder is malformed in the published abstract (0.62–0.60, which cannot bracket 0.56). We reproduce it as printed and flag it rather than quietly repairing someone else's arithmetic [78].

Both slogans are false. “DSM categories are unstable” collapses the moment anyone quotes the schizophrenia figure of 0.90. “DSM categories are stable” collapses the moment anyone quotes schizophreniform disorder at 0.29 — a label that is wrong more often than it is right. The two numbers come from the same table in the same meta-analysis. Anyone using one and not the other is arguing, not reporting.

why the same disorder scores 70% in one study and 89.2% in another

Because they are not measuring the same thing, and this is the objection a careful reader will raise, so we raise it first.

The gap between 70% and 89.2% is roughly the gap between psychiatry as practised and psychiatry done with unlimited time and a committee. Both numbers are correct. Only one of them describes the appointment you actually had.

The same cohort carries the finding that ought to worry a prescriber most. Of 1,153 patients who received a bipolar diagnosis at least once across 71,543 assessments, only 23% carried it in three-quarters or more of their assessments, and the authors report “a high prevalence of misdiagnosis and diagnostic shift from other psychiatric disorders to BD” [77]. Set that beside the bipolar-screening requirement above — the one category the FDA insists you check for before prescribing is among the ones the record is least sure about.

what to do with this. A diagnosis that changes is usually not a mistake being corrected — it is more information arriving. Bromet's authors do not conclude that diagnosis is worthless; they conclude “diagnosis should be reassessed at all follow-up points” [54]. That is a reasonable thing to expect from your own care, and a reasonable thing to ask for: when would you re-examine this diagnosis, and what would change it?

and the population picture, stated carefully

Treatment provision has risen enormously across the DSM era. Population-level distress has not fallen. Two reviews establish this and neither of them blames the manual — which is precisely why they are worth citing.

Jorm and colleagues, reviewing four countries in World Psychiatry: “the prevalence of mood and anxiety disorders and symptoms has not decreased, despite substantial increases in the provision of treatment, particularly antidepressants” [61]. Their leading explanation is not the diagnostic system but the quality and targeting of care: “much of the treatment provided does not meet the minimal standards of clinical practice guidelines and is not targeted optimally to those in greatest need” [61]. Ormel and colleagues named the pattern the treatment-prevalence paradox: “Treatments for depression have improved, and their availability has markedly increased since the 1980s. Mysteriously the general population prevalence of depression has not decreased” [62] — and attribute most of it to efficacy being overstated in the published literature and attenuated in the real world [62].

The American mortality series moves the same way. The age-adjusted suicide rate rose from 10.8 per 100,000 in 2003 to 14.2 in 2018, and “did not significantly change between 2018 and 2023 (14.1)” [63]. We give no earlier baseline because we could not retrieve one from a National Center for Health Statistics source, and we will not stitch a series together to make a longer line.

What these trends do not show. Not that treatment does not work — the trial evidence for several treatments is strong and some of it is cited above. Not that diagnosis caused anything. Not that more people are becoming ill. These are population series with dozens of inputs, and the papers that produced them explicitly decline to blame the diagnostic system. Our own work on how a definition changes a count lives at how childhood ADHD got counted — four separate series, no causal claim between any of them, and a panel deliberately left empty where the public data does not exist.

7. the case for the manual, printed whole

Everything above is the case against complacency. This section is the case for the system, made by people who know it best, and it is stronger than its critics usually allow.

the problem it solved was real, and most of it was the vocabulary

When researchers went looking for the sources of pre-1980 diagnostic disagreement, they found that the largest one was not the patients and not the clinicians. Ward and colleagues attributed disagreement to “inconstancy of the patient (5%), inconstancy of the clinician (32.5%), and inadequacy of the nomenclature (62.5%)” [66]. Nearly two thirds of the problem was the naming system itself — which is exactly the thing operational criteria fixed. Whatever else the DSM did, it addressed the documented majority cause of the failure it was built to address.

the same NIMH director, two weeks later

Insel's “Transforming Diagnosis” post is quoted everywhere. What is almost never quoted is what he co-signed a fortnight afterwards with the incoming president of the APA:

“Today, the American Psychiatric Association's (APA) Diagnostic and Statistical Manual of Mental Disorders (DSM), along with the International Classification of Diseases (ICD) represents the best information currently available for clinical diagnosis of mental disorders. Patients, families, and insurers can be confident that effective treatments are available and that the DSM is the key resource for delivering the best available care. The National Institute of Mental Health (NIMH) has not changed its position on DSM-5.”

“DSM-5 and RDoC represent complementary, not competing, frameworks… But this is a long-term undertaking. It will take years.”

Thomas Insel (NIMH) and Jeffrey Lieberman (APA), joint statement, 13 May 2013 [64]

NIMH still says the same thing about its own alternative: “RDoC is not meant to serve as a diagnostic guide, nor is it intended to replace current diagnostic systems” [65]. Anyone who quotes Insel against the DSM without quoting this is quoting half a person.

the deepest argument: no test can tell you where disorder begins

Jerome Wakefield makes the case that “the DSM has no biomarker” is not actually a coherent objection. After running through every standard validity test — statistical deviance, family history, predictive validity, factor structure, response to medication, the Robins and Guze criteria — he concludes:

“whether the distinguished constructs are disorder versus nondisorder goes beyond the test's capabilities… Thus, there is no substitute for the concept of mental disorder as the ultimate standard. None of our empirical approaches work without a warrant in a conceptual analysis of disorder.”

Jerome Wakefield, World Psychiatry 2007 [67]

A blood test can tell you a number. It cannot tell you which numbers count as illness. That judgement has to be made somewhere, by someone, and writing it down in a book that everyone can read and argue with is not obviously worse than leaving it implicit in each clinician's head.

and the honest version from a nosologist

Kenneth Kendler defends the enterprise and concedes the categories in the same paper, which is why he is worth quoting at length. The defence: “a diagnosis is real to the degree that it ‘coheres’ well with what we already know empirically and feel confident about” [69]. The concession:

“If I were to have a public debate with an arch anti-psychiatrist, I would not want to put myself in the position of defending the reality of every category in the DSM-5 or ICD-10… We have many more reasons to defend the reality of the broad classes of psychiatric illness than the specific categories in our current diagnostic manuals.”

“If we re-run the tape of history over and over again, the DSM and ICD would not likely have the same categories on every iteration.”

Kenneth Kendler, World Psychiatry 2016 [69]

Michael First adds the practical argument for not tearing it up: “Adopting a dimensional approach would likely complicate medical record keeping, create administrative and clinical barriers between mental disorders and medical conditions, require a massive retraining effort, disrupt research efforts… and complicate clinicians' efforts to integrate prior clinical research using DSM categories into clinical practice” [68].

nothing has replaced it, and the reason is the form, not the science

Every proposed successor is either not clinical, not finished, or not adopted. RDoC is disclaimed by its own creator as a research framework [65]. HiTOP's own consortium states that “to date, HiTOP has not been used clinically as a complete system” [71] — and, in the paper on integrating it into practice, its architects explain how a HiTOP clinician gets paid: by mapping onto “the appropriate ‘unspecified’ categories” of the ICD, an approach they concede “has limitations, but can provide a solution until billing and administrative procedures are better aligned” [72]. Their conclusion is the most quotable line in this whole section: “DSM and ICD codes are likely to remain the language of administrative systems for years to come” [72]. Note where they had to land: on the 62 escape hatches from section 1.

ICD-11 is the WHO's own successor, adopted in 2019 and in force since 1 January 2022. As of May 2024, by WHO's count, 14 countries and areas had begun to collect or report data using it [73]. The United States is not one of them; American claims still run on ICD-10-CM [43]. Five years after adoption, the incumbent has not moved.

The fairest summary we can write. A global survey of 1,764 clinicians in 92 countries found that the DSM and ICD earn “the highest ratings of utility for meeting administrative requirements, assigning a diagnosis, communicating with other health care professionals, and teaching” and “the lowest ratings for selecting a treatment and assessing probable prognosis” [70]. That single sentence is this entire page. The manual is very good at the job it is actually doing — being the shared name that lets care be requested, paid for, researched and taught — and its own users rate it lowest at the job everyone assumes it is doing.

8. what this means if you are the one with the diagnosis

You have just read that your label is one of hundreds of possible symptom combinations, that it was standardised in 1980 to fix a disagreement problem, that it turns into a billing code, and that its own users rate it lowest at predicting treatment. Here is what actually follows from that, and it is less than you might fear and more useful than you might expect.

Keep the diagnosis. It is doing work for you. It is what makes your care claimable, your accommodation presumable, your disability case arguable and your child's school plan triggerable. Every one of those is documented above. Nothing on this page is an argument for going without one.

five questions this page earns you

The second question is the one most worth asking, because it converts a category into something checkable. A description you can test against your own week is worth more than an explanation nobody has yet.

questions worth asking

How many symptoms do you need for a depression diagnosis?

Five of nine, present during the same two-week period, and at least one of the five must be either depressed mood or loss of interest or pleasure. The symptoms must also cause clinically significant distress or impairment in social, occupational or other important areas of functioning — that is a separate criterion, not one of the nine.

Can two people with the same diagnosis have different symptoms?

Yes, and it is not an edge case. Counting the nine depression criteria as nine yes/no items, there are 227 combinations that satisfy the diagnosis. Counting the four either/or criteria as their separate alternatives — weight up or weight down, sleeping too much or too little — there are 1,497. A 2014 study reported pairs of real patients, in real datasets, who shared the same diagnosis and no symptoms at all.

What does the DSM say it is for?

DSM-5 states that "the primary purpose of DSM-5 is to assist trained clinicians in the diagnosis of their patients' mental disorders as part of a case formulation assessment that leads to a fully informed treatment plan for each individual." The same front matter says it is "not sufficient to simply check off the symptoms in the diagnostic criteria to make a mental disorder diagnosis," and that criteria "are offered as guidelines for making diagnoses, and their use should be informed by clinical judgment."

Does the DSM tell clinicians how to treat anything?

No, and the American Psychiatric Association says so directly: "DSM does NOT include information about treatments. It is a manual for assessment and diagnosis of mental disorders and does not include information or guidelines for treatment of any disorder." The manual names conditions; it does not prescribe.

Are DSM codes what insurers bill on?

No. The APA is explicit: "There are no DSM codes." The numbers printed beside each disorder are ICD codes — ICD-10-CM in the United States since 1 October 2015 — and those are what appear on a claim. The DSM supplies the criteria that decide which code applies; the ICD supplies the code that gets paid.

Why was DSM-III in 1980 such a big deal?

Because before it, psychiatrists frequently did not agree on what a patient had. In one arm of the US–UK Diagnostic Project, 69% of American psychiatrists watching a videotaped interview said the patient had some form of schizophrenia, against 2% of the British audience. A 1974 re-analysis by Robert Spitzer and Joseph Fleiss concluded that "the reliability of psychiatric diagnosis as it has been practised since at least the late 1950s is not good." Spitzer then chaired the task force that wrote DSM-III, which replaced narrative description with explicit checkable criteria.

Does a diagnosis improve access to care?

Yes, and this is the strongest evidence-backed thing that can be said for the manual. In the APA's own words the DSM is used "in a variety of settings, including clinics, private practice, schools, hospitals, courtrooms, and the insurance industry." A coded diagnosis is what a claim needs, what a disability determination is built on, and what most school and workplace accommodation processes ask for.

Will my diagnosis still be my diagnosis in five years?

It depends enormously on which diagnosis, and that is the actual finding rather than a dodge. A meta-analysis of 42 studies and 14,484 first-episode psychosis patients, followed 4.5 years on average, ranked the categories: schizophrenia held 0.90 of the time, affective spectrum psychoses 0.84, schizoaffective disorder 0.72, delusional disorder 0.59, psychosis not otherwise specified 0.36, and schizophreniform disorder 0.29 — wrong more often than right. So "DSM diagnoses are unstable" and "DSM diagnoses are stable" are both false as blanket claims. That study covers first-episode psychosis only and says nothing about depression, ADHD or autism. A diagnosis that changes is usually not an error being corrected but more information arriving; the authors of the ten-year psychosis follow-up conclude that "diagnosis should be reassessed at all follow-up points."

Why do studies disagree about how often a diagnosis changes?

Because they measure different things, and almost nobody says so. The Spanish study reporting 70% stability for schizophrenia counted diagnoses as clinicians wrote them in 360,899 routine consultation records. The American study reporting 89.2% retention used "longitudinal best-estimate consensus diagnoses formulated after each assessment" — a research panel reviewing everything. Both figures are correct. The gap between them is roughly the gap between psychiatry as it is practised and psychiatry done with unlimited time and a committee.

Does a diagnosis improve outcomes?

That is much less clear, and we say so on the page rather than around it. The categories are reliable — different clinicians reach the same label more often than they used to — but reliability is agreement, not accuracy. The manual contains no treatment guidance, several treatments cross many categories, and the strongest defence of the system is that a shared vocabulary is a precondition for research, communication and payment at all, not that the category predicts what will help you.

sources

  1. American Psychiatric Association. "What is the DSM?" Patients & Families section, psychiatry.org. Physician review dated April 2025. Retrieved and read in full 2026-08-27. Source of the manual's stated audience and settings ("clinics, private practice, schools, hospitals, courtrooms, and the insurance industry"), the "common language" purpose, the statement that "DSM does NOT include information about treatments", the three-component description, and the "DSM, ICD and insurance" passage. Note that psychiatry.org returns HTTP 403 to some automated clients; this page was retrieved with a standard browser user-agent and the quoted text read from the returned HTML. https://www.psychiatry.org/patients-families/what-is-the-dsm
  2. American Psychiatric Association. "About DSM-5-TR," psychiatry.org. Retrieved 2026-08-27. Source of "DSM consists of three major components: the diagnostic classification, the diagnostic criteria sets, and the descriptive text", of the statement that each diagnosis carries a code "typically used by individual providers, institutions, and agencies for data collection and billing purposes" derived from ICD-10-CM, and of the caution that the criteria "are meant to be used by trained professionals using clinical judgment; they are not meant to be used by the general public." https://www.psychiatry.org/psychiatrists/practice/dsm/about-dsm
  3. American Psychiatric Association. "DSM History," psychiatry.org. Retrieved 2026-08-27. Source of the 1840 census "idiocy/insanity" entry, the seven 1880 census categories, the 1917 uniform-statistics plan, the post-war Army classification "modified by the Veterans Administration", the ICD-6 description ("10 categories for psychoses and psychoneuroses and seven categories for disorders of character, behavior, and intelligence"), and the DSM-III innovations passage. Two cautions we print rather than hide: the APA page names the 1917 partner "National Commission on Mental Hygiene", where the historical record and Shorter [18] give the National Committee for Mental Hygiene; and the APA's ICD-6 category counts are rendered differently by several secondary sources, so we quote the APA's own wording rather than a total. https://www.psychiatry.org/psychiatrists/practice/dsm/about-dsm/history-of-the-dsm
  4. American Psychiatric Association. "Insurance Implications of DSM-5," 4 pp., © 2013 American Psychiatric Association. PDF retrieved from the psychiatry.org File Library and read in full 2026-08-27. Source of "developed to facilitate a seamless transition into immediate use by clinicians and insurers", the HIPAA-approved-coding-system passage, the statement that the GAF "was used for determinations of medical necessity for treatment by many payers, and eligibility for short- and long-term disability compensation", the shared-code examples (hoarding disorder and OCD both coding to ICD-10-CM F42; DMDD to F34.8), and "Clinicians use DSM-5 diagnoses to communicate with their patients and with other clinicians, and to request reimbursement from insurance organizations." One date in this document was overtaken by events and we say so on the page: it gives 1 October 2014 as the ICD-10-CM adoption date, which Congress delayed; CMS records the actual transition as 1 October 2015 [33]. https://www.psychiatry.org/File%20Library/Psychiatrists/Practice/DSM/APA_DSM_Insurance-Implications-of-DSM-5.pdf
  5. American Psychiatric Association. "DSM-5 Table of Contents," 8 pp. PDF retrieved from the psychiatry.org File Library 2026-08-27. Source of the manual's three-section structure — Section I (Introduction, Use of the Manual, Cautionary Statement for Forensic Use of DSM-5), Section II (Diagnostic Criteria and Codes), Section III (Emerging Measures and Models, containing the Alternative DSM-5 Model for Personality Disorders and the Conditions for Further Study). The residual-category count on this page is our own count from this document: 24 entries beginning "Other Specified" and 38 beginning "Unspecified", 62 in total. The method is a line-level match on the published classification list; anyone can reproduce it from the same PDF. The APA publishes no count of its own. https://www.psychiatry.org/File%20Library/Psychiatrists/Practice/DSM/APA_DSM-5-Contents.pdf
  6. American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5). Arlington, VA: American Psychiatric Publishing, 2013. HOW THESE QUOTATIONS WERE VERIFIED, because it matters: the APA's own DSM Library (dsm.psychiatryonline.org) is paywalled and refuses automated retrieval, so no quotation here was taken from an APA server. Each sentence was instead confirmed by exact-phrase search inside the scanned volumes held in the Internet Archive / Open Library full-text index — catalogue records "diagnosticstatis0000amer" and "diagnosticstatis0005unse", both DSM-5, plus APA Publishing's own Desk Reference to the Diagnostic Criteria From DSM-5 ("deskreferencetod0000unse"). Every phrase quoted on this page returned a hit inside at least one of those three volumes on 2026-08-27; control searches on altered wording returned zero, which is what establishes that the index matches phrases exactly rather than approximately. Two of the quotations — the definition of a mental disorder at [7] and the impairment-and-disability clause in section 2 — matched only inside APA Publishing's Desk Reference and Guidebook rather than the DSM-5 scans themselves, most likely because of line-break hyphenation in the scanned page; both are APA-published volumes quoting the APA manual, and the page says so here rather than letting a bare citation imply otherwise. EDITION WARNING: all of it is DSM-5 (2013). The current edition is DSM-5-TR (2022), whose front matter we could not verify against an authorised copy, and the page never labels a DSM-5 sentence as DSM-5-TR. https://openlibrary.org/search/inside?q=%22Hence%2C+it+is+not+sufficient+to+simply+check+off+the+symptoms+in+the+diagnostic+criteria+to+make+a+mental+disorder+diagnosis%22
  7. Black DW, Grant JE. DSM-5 Guidebook: The Essential Companion to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition. American Psychiatric Publishing, 2014. Used as the carrier for the DSM-5 definition of a mental disorder, which returned an exact-phrase hit in this APA-published companion volume in the same full-text index described at [6]. An APA Publishing title quoting the APA manual is the closest thing to a primary source that is publicly checkable here. https://openlibrary.org/search/inside?q=%22A+mental+disorder+is+a+syndrome+characterized+by+clinically+significant+disturbance+in+an+individual%27s+cognition%2C+emotion+regulation%2C+or+behavior%22
  8. Maji S, Mohapatra D, Jena M, Srinivasan A, Maiti R. "Repurposing of dextromethorphan as an adjunct therapy in patients with major depressive disorder: a systematic review and meta-analysis." BMJ Open 2024;14:e080500. doi:10.1136/bmjopen-2023-080500. PMC11086520, open access, full text read 2026-08-27. Used here only because it block-quotes the DSM-5 major depressive disorder criteria: Criterion A ("Five (or more) of the following symptoms have been present during the same 2-week period... at least one of the symptoms is either depressed mood or loss of interest or pleasure"), Criterion B (the clinically-significant-distress-or-impairment clause) and Criterion C. The quotation renders the criteria as prose and omits the APA's parenthetical numbering, which is why the structure table on this page also cites [9]. https://doi.org/10.1136/bmjopen-2023-080500
  9. Substance Abuse and Mental Health Services Administration. DSM-5 Changes: Implications for Child Serious Emotional Disturbance. Rockville (MD): SAMHSA, June 2016. Table 9, "DSM-IV to DSM-5 Major Depressive Episode/Disorder Comparison." NCBI Bookshelf NBK519712, US government publication, public domain, read 2026-08-27. Corroborates the five-of-nine structure ("Five or more of the following A Criteria (at least one includes A1 or A2)") and the impairment clause. Note that SAMHSA operationalises the duration as "most of the day, daily, for at least 2 weeks in a row", which is the table's working restatement and not the APA's own wording — the page quotes APA's "during the same 2-week period" for that reason. https://www.ncbi.nlm.nih.gov/books/NBK519712/table/ch3.t5/
  10. Fried EI, Nesse RM. "The impact of individual depressive symptoms on impairment of psychosocial functioning." PLoS ONE 2014;9(2):e90311. doi:10.1371/journal.pone.0090311. PMC3938686, open access, full text read 2026-08-27. The source we actually quote for the 1,497 figure, because the upstream letter [11] is unreadable: "Four symptoms are compound symptoms comprised by different subsymptoms... leading to 1,497 unique symptom profiles that all qualify for the same diagnosis, including profiles that do not have a single symptom in common." https://doi.org/10.1371/journal.pone.0090311
  11. Østergaard SD, Jensen SOW, Bech P. "The heterogeneity of the depressive syndrome: when numbers get serious." Acta Psychiatrica Scandinavica 2011;124(6):495–496. doi:10.1111/j.1600-0447.2011.01744.x. PMID 21838736. This is the origin of the 1,497 figure and we could not read one word of it: it is a two-page letter with no abstract in PubMed, none in Europe PMC, none deposited with Crossref, and the publisher PDF returns HTTP 403. It is cited here only as the upstream reference identified inside Fried & Nesse 2014 [10], which is where the number is quoted from. It was computed on DSM-IV criteria, whose nine-symptom A-list is structurally identical to DSM-5's. https://doi.org/10.1111/j.1600-0447.2011.01744.x
  12. Zimmerman M, Ellison W, Young D, Chelminski I, Dalrymple K. "How many different ways do patients meet the diagnostic criteria for major depressive disorder?" Comprehensive Psychiatry 2015;56:29–34. doi:10.1016/j.comppsych.2014.09.007. PMID 25266848. Closed access; quoted from the publisher abstract only, retrieved 2026-08-27. Source of "There are 227 possible ways to meet the symptom criteria for major depressive disorder (MDD)", of the finding that patients "met the MDD symptom criteria in 170 different ways", and of "one-quarter (57/227) of the criteria combinations did not occur." The sample is described in the abstract only as "more than 1500 patients"; the exact number is not public and is not stated on this page. https://doi.org/10.1016/j.comppsych.2014.09.007
  13. Galatzer-Levy IR, Bryant RA. "636,120 Ways to Have Posttraumatic Stress Disorder." Perspectives on Psychological Science 2013;8(6):651–662. doi:10.1177/1745691613504115. Quoted from the publisher-deposited abstract retrieved via Crossref 2026-08-27: "we demonstrate that the DSM–IV criteria listed for PTSD have a high level of symptom profile heterogeneity (79,794 combinations); the changes result in an eightfold expansion in the DSM–5, to 636,120 combinations." Both figures reproduce exactly under the criteria structure they describe. https://doi.org/10.1177/1745691613504115
  14. Olbert CM, Gala GJ, Tupler LA. "Quantifying heterogeneity attributable to polythetic diagnostic criteria: theoretical framework and empirical application." Journal of Abnormal Psychology 2014;123(2):452–462. doi:10.1037/a0036068. PMID 24886017. Closed access; quoted from the abstract only. Source of "in most categories, 2 individuals with the same diagnosis may share no symptoms in common" across 18 diagnostic categories, and of the empirical finding in two large datasets that "For both disorders in each of the datasets, pairs of individuals who shared no common symptoms were observed." The abstract states no direction of change between DSM-IV and DSM-5, so this page makes no such claim. https://doi.org/10.1037/a0036068
  15. Fried EI, Nesse RM. "Depression is not a consistent syndrome: An investigation of unique symptom patterns in the STAR*D study." Journal of Affective Disorders 2015;172:96–102. doi:10.1016/j.jad.2014.10.010. PMC4397113, full text read 2026-08-27. Source of "we identified 1030 unique symptom profiles" among 3,703 patients, of "501 profiles (48.6%) were endorsed by only one individual", and of the patient-level restatement in the discussion, "roughly 14% of the participants exhibited unique profiles not shared with a single other person in the study." Three qualifications this page prints rather than omits: the analysis used twelve QIDS items, not the nine DSM criteria; profiles were not required to include a DSM core symptom; and 48.6% is a share of profiles, not of patients. https://doi.org/10.1016/j.jad.2014.10.010
  16. Buss JF, Watts AL, Lorenzo-Luaces L. "Methods for quantifying the heterogeneity of psychopathology." BMC Psychiatry 2023;23:897. doi:10.1186/s12888-023-05377-5. PMC10690966, open access, read 2026-08-27. Cited to show the finding is current rather than a decade-old curiosity: "Using polythetic criteria for an MDE leads to highly heterogeneous symptom presentations to the point that two individuals with an MDE may not share a single symptom." https://doi.org/10.1186/s12888-023-05377-5
  17. Surís A, Holliday R, North CS. "The Evolution of the Classification of Psychiatric Disorders." Behavioral Sciences 2016;6(1):5. doi:10.3390/bs6010005. PMC4810039, open access. Source of the Statistical Manual for the Use of Institutions for the Insane ("21 disorders, 19 of which were psychotic disorders... across a series of 10 editions by 1942"), of the five competing American classifications in the 1950s (asylums, Army, Navy, VA, American Prison Association), of DSM-I being "based on the Veterans Administration (VA) system", of the diagnosis-count series and of the page-count growth "from 130 pages in the first edition" to "992 pages in DSM-5." https://doi.org/10.3390/bs6010005
  18. Shorter E. "The history of nosology and the rise of the Diagnostic and Statistical Manual of Mental Disorders." Dialogues in Clinical Neuroscience 2015;17(1):59–67. PMC4421901, open access. Source of the Medical 203 passage: the nosology "directed by psychoanalyst William Menninger, who, during World War II, was a brigadier-general and the head of psychiatry in the Office of the Surgeon General", appearing "in October 1945 as the Technical Medical Bulletin number 203 of the United States Army." Also the source of the dissent this page prints on how much the Research Diagnostic Criteria actually shaped DSM-III. https://pmc.ncbi.nlm.nih.gov/articles/PMC4421901/
  19. Houts AC. "Fifty years of psychiatric nomenclature: reflections on the 1943 War Department Technical Bulletin, Medical 203." Journal of Clinical Psychology 2000;56(7):935–967. PMID 10902952. Abstract only. "The history of Medical 203 is presented to show how Medical 203 was adapted to become DSM-I." The companion reprint of the bulletin itself is J Clin Psychol 2000;56(7):925–934, PMID 10902951. THE DATE IS CONTESTED AND THIS PAGE SAYS SO: Houts' title says 1943, the reprint record says 1946, Shorter [18] says October 1945, and the APA files the document under "Post–World War II" [3]. We give the range rather than pick one. https://pubmed.ncbi.nlm.nih.gov/10902952/
  20. Kawa S, Giordano J. "A brief historicity of the Diagnostic and Statistical Manual of Mental Disorders: issues and implications for the future of psychiatric canon and practice." Philosophy, Ethics, and Humanities in Medicine 2012;7:2. doi:10.1186/1747-5341-7-2. PMC3282636, open access. Source of the diagnosis counts "from 182 in the DSM-II to 265 in the DSM-III", of the removal and partial reinstatement of "neurosis", and of the funding-and-legitimacy passage: "Diminished research support from the National Institute of Mental Health (NIMH), and reduced resource allocations by the federal government and insurance providers were prevalent throughout the 1970s." https://doi.org/10.1186/1747-5341-7-2
  21. Kirk SA, Kutchins H. "The Myth of the Reliability of DSM." The Journal of Mind and Behavior 1994;15(1&2):71–86. Reprinted in full with the journal's and the authors' permission by the Academy of Analytic Arts; full text read 2026-08-27. The carrier for the Spitzer & Fleiss quotations at page-level ([22]) and the source of the critique this page prints beside them — that the same kappa table "could have described the state of reliability as quite good", that "in every single diagnostic category in every study, psychiatric agreement was considerably better than chance", and that the paper's senior author was appointed chair of the DSM-III Task Force being formed as it appeared. https://academyanalyticarts.org/kirk-myth-reliability-dsm
  22. Spitzer RL, Fleiss JL. "A re-analysis of the reliability of psychiatric diagnosis." British Journal of Psychiatry 1974;125:341–347. doi:10.1192/bjp.125.4.341. PMID 4425771. No open copy exists and the publisher platform was serving a service-suspension notice when we checked on 2026-08-27, so the two sentences quoted on this page — "There are no diagnostic categories for which reliability is uniformly high..." (p. 344) and "The reliability of psychiatric diagnosis as it has been practised since at least the late 1950s is not good" (p. 345) — are quoted AS QUOTED, with page numbers, in Kirk & Kutchins 1994 [21]. That is stated on the page rather than hidden behind a bare citation. https://doi.org/10.1192/bjp.125.4.341
  23. Feighner JP, Robins E, Guze SB, Woodruff RA Jr, Winokur G, Munoz R. "Diagnostic criteria for use in psychiatric research." Archives of General Psychiatry 1972;26(1):57–63. doi:10.1001/archpsyc.1972.01750190059011. PMID 5009428. Closed access. The category count is contested in the peer-reviewed literature — Escobar 2013 [32] and Dean 2017 [28] both say 14, and Surís 2016 [17] enumerates a list that reconciles to 14, while at least one later paper says 20. This page gives 14 and names the disagreement. https://doi.org/10.1001/archpsyc.1972.01750190059011
  24. Robins E, Guze SB. "Establishment of diagnostic validity in psychiatric illness: its application to schizophrenia." American Journal of Psychiatry 1970;126(7):983–987. doi:10.1176/ajp.126.7.983. PMID 5409569. Closed access; the five-phase framework and the "continuing self-rectification" quotation are taken from Rodríguez Puente A, "The DSM validation method and its decision points," History and Philosophy of the Life Sciences 2026;48, PMC13083308, open access, read 2026-08-27. https://pmc.ncbi.nlm.nih.gov/articles/PMC13083308/
  25. Spitzer RL, Endicott J, Robins E. "Research diagnostic criteria: rationale and reliability." Archives of General Psychiatry 1978;35(6):773–782. doi:10.1001/archpsyc.1978.01770300115013. PMID 655775. Abstract read 2026-08-27: "A crucial problem in psychiatry, affecting clinical work as well as research, is the generally low reliability of current psychiatric diagnostic procedures." https://doi.org/10.1001/archpsyc.1978.01770300115013
  26. Freeborn A. "Testing psychiatrists to diagnose schizophrenia: Crisis, consensus, and computers in post-war psychiatry." History of the Human Sciences 2025. doi:10.1177/09526951241309504. Author copy read in full 2026-08-27 from the Max Planck Institute repository. Source of the videotape result — "After watching 40 minutes of video of Patient F being interviewed, 69% of the Americans said the patient had some form of schizophrenia, compared with only 2% of the UK audience" — which Freeborn attributes to Kendell et al. 1971, and of Spitzer's own account to David Healy that the US–UK study "inevitably led to justifying diagnostic criteria." https://doi.org/10.1177/09526951241309504
  27. Mayes R, Horwitz AV. "DSM-III and the revolution in the classification of mental illness." Journal of the History of the Behavioral Sciences 2005;41(3):249–267. doi:10.1002/jhbs.20103. PMID 15981242. Closed access; quoted from the publisher abstract, which is itself the argument: the shift "was neither a product of growing scientific knowledge nor of increasing medicalization", and the four listed drivers include "mounting pressure on psychiatrists from health insurers to demonstrate the effectiveness of their practices" and "the necessity of pharmaceutical companies to market their products to treat specific diseases." https://doi.org/10.1002/jhbs.20103
  28. Dean CE. "Social inequality, scientific inequality, and the future of mental illness." Philosophy, Ethics, and Humanities in Medicine 2017;12:10. doi:10.1186/s13010-017-0052-x. PMC5738232, open access. Source of the regulatory driver — "a 1962 amendment to the Food, Drugs, and Cosmetics Act requiring that prescription drugs were to be used for the treatment of categorical (specific) diseases" — and of the DSM-III foreword's own disclaimer that "there is no assumption that each mental disorder is a discrete entity with sharp boundaries." https://doi.org/10.1186/s13010-017-0052-x
  29. Scull A. "Rosenhan revisited: successful scientific fraud." History of Psychiatry 2023;34(2):180–195. doi:10.1177/0957154X221150878. PMID 36737877. Abstract read 2026-08-27. Peer-reviewed, by a historian of psychiatry working from records shared by the journalist Susannah Cahalan: "based on the findings of an investigative journalist, Susannah Cahalan, and on records she shared with the author, we now know that this research is a spectacularly successful case of scientific fraud." This page cites Rosenhan [30] nowhere without this beside it. https://doi.org/10.1177/0957154X221150878
  30. Rosenhan DL. "On being sane in insane places." Science 1973;179(4070):250–258. doi:10.1126/science.179.4070.250. PMID 4683124. Cited only as the object of [29] and [31]. https://doi.org/10.1126/science.179.4070.250
  31. Spitzer RL. "More on pseudoscience in science and the case for psychiatric diagnosis. A critique of D.L. Rosenhan's 'On Being Sane in Insane Places' and 'The Contextual Nature of Psychiatric Diagnosis'." Archives of General Psychiatry 1976;33(4):459–470. PMID 938183. Abstract read 2026-08-27: "Rosenhan's 1973 article... was pseudoscience presented as science... a careful examination of this study's methods, results, and conclusions leads to a diagnosis of 'logic in remission'." The earlier critique is Spitzer RL, J Abnorm Psychol 1975;84(5):442–452, PMID 1194504, which carries no abstract. https://pubmed.ncbi.nlm.nih.gov/938183/
  32. Escobar JI. "An insider's view of the new diagnostic and statistical manual of North American psychiatry (DSM-5)." Colombia Médica 2013;44(3). PMC4002019, open access. Escobar served on a DSM-5 work group, which is why this page treats him as an insider: "DSM-III started placing particular emphasis on diagnostic reliability or agreement among observers more than on validity", and "as DSM-III and IV criteria resulted from a strong desire to increase reliability (agreement across evaluators), they sacrificed validity." https://pmc.ncbi.nlm.nih.gov/articles/PMC4002019/
  33. Centers for Medicare & Medicaid Services. "ICD-10 codes," cms.gov. Retrieved 2026-08-27. "On October 1, 2015, the health care industry transitioned from ICD-9 to ICD-10 codes for diagnoses and inpatient procedures... ICD-10 applies to all parties covered by the Health Insurance Portability and Accountability Act (HIPAA), not just providers who bill Medicare." https://www.cms.gov/medicare/coding-billing/icd-10-codes
  34. American Psychiatric Association. "DSM Frequently Asked Questions," psychiatry.org. Retrieved and read 2026-08-27. Source of the flat statement "There are no DSM codes. The codes appearing in DSM-5 in addition to the ICD-10-CM codes were ICD-9 codes. These codes are no longer valid since ICD-10-CM codes replaced them in the United States starting October 1, 2015. Clinicians should use ICD-10-CM codes to submit claims", and of "DSM-5-TR, like DSM-5, is a manual for assessment and diagnosis of mental disorders and does not include information or guidelines for treatment of any disorder." https://www.psychiatry.org/psychiatrists/practice/dsm/frequently-asked-questions
  35. Social Security Administration. Listing of Impairments, 20 CFR Part 404, Subpart P, Appendix 1, section 12.00 "Mental Disorders." Read from the codified text on govinfo 2026-08-27. Source of the eleven listed categories and of "We need objective medical evidence from an acceptable medical source to establish that you have a medically determinable mental disorder." A finding we print because it cuts against the easy version of the access argument: a full-text search of Appendix 1 returns ZERO occurrences of "DSM" or "Diagnostic and Statistical." The codified listing never names the manual. https://www.govinfo.gov/content/pkg/CFR-2024-title20-vol2/xml/CFR-2024-title20-vol2-part404-subpartP-app1.xml
  36. Social Security Administration. "Revised Medical Criteria for Evaluating Mental Disorders," final rule, 81 FR 66138, 26 September 2016. Federal Register full text read 2026-08-27. The DSM link and the disclaimer both live here rather than in the codified listing: "We revised most of the paragraph A criteria using the diagnostic features for the corresponding categories of mental disorders in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5)" and "Although a claimant must have a medically determinable mental impairment, the claimant does not have to have a diagnosis for his or her mental impairment to satisfy the listing criteria." Also the source of "our mental disorders listings are function-driven, not diagnosis-driven." https://www.federalregister.gov/documents/full_text/text/2016/09/26/2016-22908.txt
  37. Social Security Administration. 20 CFR 404.1521, "Establishing that you have a medically determinable impairment(s)." Codified text read on govinfo 2026-08-27: "a physical or mental impairment must be established by objective medical evidence from an acceptable medical source. We will not use your statement of symptoms, a diagnosis, or a medical opinion to establish the existence of an impairment(s)." https://www.govinfo.gov/content/pkg/CFR-2024-title20-vol2/xml/CFR-2024-title20-vol2-sec404-1521.xml
  38. US Department of Education. 34 CFR 300.8, "Child with a disability" (IDEA Part B regulations). Codified text read on govinfo 2026-08-27. Source of the "emotional disturbance" definition at 300.8(c)(4) and the "other health impairment" definition at 300.8(c)(9), which names "attention deficit disorder or attention deficit hyperactivity disorder" in the regulation itself. Neither definition mentions diagnosis, DSM, ICD, physician or clinician anywhere. Eligibility procedure: 34 CFR 300.306(a)(1) (team determination), 300.304(b)(2) ("Not use any single measure or assessment as the sole criterion"). https://www.govinfo.gov/content/pkg/CFR-2023-title34-vol2/xml/CFR-2023-title34-vol2-sec300-8.xml
  39. US Department of Education. IDEA Part B final regulations, preamble, 71 FR 46540 at 46551, 14 August 2006. Federal Register full text read 2026-08-27: "there is nothing in the Act that would prevent a State from requiring a medical evaluation for eligibility under other health impairment, provided the medical evaluation is conducted at no cost to the parent." https://www.federalregister.gov/documents/full_text/text/2006/08/14/06-6656.txt
  40. US Department of Education, Office for Civil Rights. "Dear Colleague Letter and Resource Guide on Students with ADHD," 26 July 2016. PDF read 2026-08-27. Source of both halves of the Section 504 position: "there is nothing in Section 504 that requires a medical assessment as a precondition to the school district's determination that the student has a disability" (p. 23) and footnote 70, "A specific diagnosis is not actually necessary if the school determines a student is substantially limited in a major life activity"; and, on the other side, "OCR will presume, unless there is evidence to the contrary, that a student with a diagnosis of ADHD is substantially limited in one or more major life activities" (p. 15). That presumption is the single clearest documented statement of what a diagnosis buys a family. https://www.ed.gov/about/offices/list/ocr/letters/colleague-201607-504-adhd.pdf
  41. Equal Employment Opportunity Commission. 29 CFR 1630.2, ADA regulations. Codified text read on govinfo 2026-08-27. 1630.2(h)(2) defines mental impairment functionally, but 1630.2(j)(3)(iii) — the "predictable assessments" rule — hard-codes DSM names: "major depressive disorder, bipolar disorder, post-traumatic stress disorder, obsessive compulsive disorder, and schizophrenia substantially limit brain function." https://www.govinfo.gov/content/pkg/CFR-2024-title29-vol4/xml/CFR-2024-title29-vol4-sec1630-2.xml
  42. Equal Employment Opportunity Commission. "Depression, PTSD, & Other Mental Health Conditions in the Workplace: Your Legal Rights," EEOC-NVTA-2016-11, issued 12 December 2016. Read 2026-08-27: "Mental health conditions like major depression, post-traumatic stress disorder (PTSD), bipolar disorder, schizophrenia, and obsessive compulsive disorder (OCD) should easily qualify" and, on documentation, "If you do not want the employer to know your specific diagnosis, it may be enough to provide documentation that describes your condition more generally (by stating, for example, that you have an 'anxiety disorder')." https://www.eeoc.gov/laws/guidance/depression-ptsd-other-mental-health-conditions-workplace-your-legal-rights
  43. US Department of Health and Human Services. 45 CFR 162.1002(c)(2), HIPAA standard code sets. Codified text read on govinfo 2026-08-27: "For the period on and after October 1, 2015... International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM)... for the following conditions: (i) Diseases. (ii) Injuries. (iii) Impairments..." This is the regulation that makes ICD-10-CM the code set every covered claim must speak. https://www.govinfo.gov/content/pkg/CFR-2024-title45-vol2/xml/CFR-2024-title45-vol2-part162.xml
  44. Centers for Medicare & Medicaid Services. 42 CFR 424.32(a)(2): "A claim for physician services, clinical psychologist services, or clinical social worker services must include appropriate diagnostic coding for those services." Read 2026-08-27. Note that the codified text still names ICD-9-CM, an un-updated cross-reference superseded by 45 CFR 162.1002 [43]; this page quotes the requirement, not the obsolete code-set name. https://www.ecfr.gov/current/title-42/chapter-IV/subchapter-B/part-424/subpart-C/section-424.32
  45. National Institute of Mental Health. "Mental Illness" statistics page, reporting 2022 National Survey on Drug Use and Health. Read 2026-08-27: 23.1% of US adults had any mental illness (59.3 million), and "among the 59.3 million adults with AMI, 30.0 million (50.6%) received mental health treatment in the past year." https://www.nimh.nih.gov/health/statistics/mental-illness
  46. Wang PS, Lane M, Olfson M, Pincus HA, Wells KB, Kessler RC. "Twelve-month use of mental health services in the United States: results from the National Comorbidity Survey Replication." Archives of General Psychiatry 2005;62(6):629–640. doi:10.1001/archpsyc.62.6.629. "Of 12-month cases, 41.1% received some treatment in the past 12 months"; "Most people with mental disorders in the United States remain either untreated or poorly treated." https://doi.org/10.1001/archpsyc.62.6.629
  47. Kvaale EP, Haslam N, Gottdiener WH. "The 'side effects' of medicalization: a meta-analytic review of how biogenetic explanations affect stigma." Clinical Psychology Review 2013;33(6):782–794. doi:10.1016/j.cpr.2013.06.002. Meta-analysis of 28 experimental studies: "biogenetic explanations reduce blame (Hedges g = −0.324) but induce pessimism (Hedges g = 0.263)... increase endorsement of the stereotype that people with psychological problems are dangerous (Hedges g = 0.198), although this result could reflect publication bias." The authors' own publication-bias caveat is printed on this page with the figure. https://doi.org/10.1016/j.cpr.2013.06.002
  48. "Exploring Lived Experiences of Receiving a Diagnosis of Autism in Adulthood: A Systematic Review." Autism in Adulthood 2026;7:1–12. doi:10.1089/aut.2023.0152. Thematic synthesis of 26 studies; conclusion quoted verbatim: "While diagnosis can grant understanding, acceptance, and community, autistic adults continue to struggle with stigma, regret, and inadequate support post-diagnosis." https://doi.org/10.1089/aut.2023.0152
  49. Insel T. "Transforming Diagnosis." NIMH Director's Blog, 29 April 2013. The live NIMH URL 404s; quoted from the Internet Archive capture of 3 May 2013, read 2026-08-27. "The strength of each of the editions of DSM has been 'reliability'... The weakness is its lack of validity"; "symptoms alone rarely indicate the best choice of treatment"; "RDoC, for now, is a research framework, not a clinical tool." http://web.archive.org/web/20130503200029/http://www.nimh.nih.gov/about/director/2013/transforming-diagnosis.shtml
  50. Hyman SE. "Diagnosing the DSM: Diagnostic Classification Needs Fundamental Reform." Cerebrum, 26 April 2011. PMC3574782, open access, read 2026-08-27. Used in preference to Hyman's closed-access 2010 Annual Review paper. Source of "the DSM-IV is often treated more like the periodic table of elements than as a highly useful but limited product of expert committees", of "the de facto reification of its diagnostic silos", and of "the DSM-IV handling of each disorder as a discrete natural category, discontinuous from other categories of disorder and from health, is palpably wrong." Hyman wrote this as a member of the DSM-5 Task Force and chair of the ICD-11 advisory group, which he discloses in the article — which is exactly why this page treats him as an insider. https://pmc.ncbi.nlm.nih.gov/articles/PMC3574782/
  51. Kessler RC, Chiu WT, Demler O, Walters EE. "Prevalence, severity, and comorbidity of twelve-month DSM-IV disorders in the National Comorbidity Survey Replication." Archives of General Psychiatry 2005;62(6):617–627. doi:10.1001/archpsyc.62.6.617. PMC2847357. Source of "55% carry only a single diagnosis, 22% two, and 23% three or more" — the 45% figure on this page is 22 + 23 and is stated as such. https://doi.org/10.1001/archpsyc.62.6.617
  52. Caspi A, Houts RM, Belsky DW, et al. "The p Factor: One General Psychopathology Factor in the Structure of Psychiatric Disorders?" Clinical Psychological Science 2014;2(2):119–137. doi:10.1177/2167702613497473. PMC4209412. "The p factor explains why it is challenging to find causes, consequences, biomarkers, and treatments with specificity to individual mental disorders." https://doi.org/10.1177/2167702613497473
  53. Baca-Garcia E, Perez-Rodriguez MM, Basurte-Villamor I, et al. "Diagnostic stability of psychiatric disorders in clinical practice." British Journal of Psychiatry 2007;190:210–216. doi:10.1192/bjp.bp.106.024026. 10,025 adults, each assessed at least ten times across 360,899 consultations. "The temporal consistency of mental disorders was poor, ranging from 29% for specific personality disorders to 70% for schizophrenia, with stability greatest for in-patient diagnoses and least for out-patient diagnoses. The findings are an indictment of our current psychiatric diagnostic practice." https://doi.org/10.1192/bjp.bp.106.024026
  54. Bromet EJ, Kotov R, Fochtmann LJ, et al. "Diagnostic shifts during the decade following first admission for psychosis." American Journal of Psychiatry 2011;168(11):1186–1194. doi:10.1176/appi.ajp.2011.11010048. PMC3589618. 470 first-admission psychosis patients followed ten years: "diagnoses were changed for 50.7% of study participants at some point during the study" — but also "Most participants who were initially diagnosed with schizophrenia or bipolar disorder retained the diagnosis at year 10 (89.2% and 77.8%, respectively)." Both halves are quoted on this page. https://doi.org/10.1176/appi.ajp.2011.11010048
  55. American Psychiatric Association. "Highlights of Changes from DSM-IV-TR to DSM-5." PDF retrieved from the psychiatry.org File Library. The APA in its own voice on why a set of its own categories was deleted: the DSM-IV schizophrenia subtypes were eliminated "due to their limited diagnostic stability, low reliability, and poor validity. These subtypes also have not been shown to exhibit distinctive patterns of treatment response or longitudinal course." Also the source of "An inability to find scientifically meaningful differences between these two conditions led to their combination" on dysthymia and chronic major depression. https://www.psychiatry.org/File%20Library/Psychiatrists/Practice/DSM/APA_DSM_Changes_from_DSM-IV-TR_-to_DSM-5.pdf
  56. US Food and Drug Administration. Sertraline hydrochloride prescribing information, retrieved via the openFDA drug label API 2026-08-27. Source of the six DSM-anchored indications (major depressive disorder, obsessive-compulsive disorder, panic disorder, PTSD, premenstrual dysphoric disorder, social anxiety disorder), each tied in the label text to a named DSM edition; and of the bipolar-screening warning quoted on this page, including the FDA's own hedge: "It is generally believed (though not established in controlled trials) that treating such an episode with an antidepressant alone may increase the likelihood of precipitation of a mixed/manic episode." The quetiapine fumarate label, retrieved the same way, carries indications for schizophrenia, bipolar I manic episodes and bipolar depressive episodes. https://api.fda.gov/drug/label.json?search=openfda.generic_name:%22sertraline+hydrochloride%22
  57. Barlow DH, Farchione TJ, Bullis JR, et al. "The Unified Protocol for Transdiagnostic Treatment of Emotional Disorders Compared With Diagnosis-Specific Protocols for Anxiety Disorders: A Randomized Clinical Trial." JAMA Psychiatry 2017;74(9):875–884. doi:10.1001/jamapsychiatry.2017.2164. PMC5710228. 223 patients with panic disorder, generalised anxiety disorder, OCD or social anxiety disorder: results "indicated statistical equivalence between the UP and SDPs", and completion was more likely on the transdiagnostic protocol (OR 3.11; 95% CI 1.44–6.74). https://doi.org/10.1001/jamapsychiatry.2017.2164
  58. Cipriani A, Hawton K, Stockton S, Geddes JR. "Lithium in the prevention of suicide in mood disorders: updated systematic review and meta-analysis." BMJ 2013;346:f3646. doi:10.1136/bmj.f3646. 48 randomised trials, 6,674 participants. "Lithium was more effective than placebo in reducing the number of suicides (odds ratio 0.13, 95% CI 0.03 to 0.66)." The finding this page prints beside it, from the same paper: "In unipolar depression, lithium was associated with a reduced risk of suicide (0.36, 0.13 to 0.98)" — the archetypal bipolar drug is not confined to the bipolar category. https://doi.org/10.1136/bmj.f3646
  59. Cortese S, Adamo N, Del Giovane C, et al. "Comparative efficacy and tolerability of medications for attention-deficit hyperactivity disorder in children, adolescents, and adults: a systematic review and network meta-analysis." Lancet Psychiatry 2018;5(9):727–738. doi:10.1016/S2215-0366(18)30269-4. 133 double-blind randomised trials. In children and adolescents on clinician ratings: amphetamines SMD −1.02 (95% CI −1.19 to −0.85), methylphenidate −0.78 (−0.93 to −0.62), atomoxetine −0.56 (−0.66 to −0.45). The authors' own caveat, printed here: on teachers' ratings only methylphenidate reached significance, and confidence in the estimates ranged from high down to very low for most indirect comparisons. https://doi.org/10.1016/S2215-0366(18)30269-4
  60. Penttilä M, Jääskeläinen E, Hirvonen N, Isohanni M, Miettunen J. "Duration of untreated psychosis as predictor of long-term outcome in schizophrenia: systematic review and meta-analysis." British Journal of Psychiatry 2014;205(2):88–94. doi:10.1192/bjp.bp.113.127753. 33 samples: long duration of untreated psychosis correlated with poorer symptomatic, social and global outcome, "correlations 0.13–0.18", and the authors describe this as "a small but mostly consistent correlation" indicating early intervention "may have at least subtle positive effects." The magnitude is stated on this page rather than rounded up. https://doi.org/10.1192/bjp.bp.113.127753
  61. Jorm AF, Patten SB, Brugha TS, Mojtabai R. "Has increased provision of treatment reduced the prevalence of common mental disorders? Review of the evidence from four countries." World Psychiatry 2017;16(1):90–99. doi:10.1002/wps.20388. PMC5269479, open access. "the prevalence of mood and anxiety disorders and symptoms has not decreased, despite substantial increases in the provision of treatment, particularly antidepressants." Critically, the authors' leading explanation is treatment quality and targeting — "much of the treatment provided does not meet the minimal standards of clinical practice guidelines and is not targeted optimally to those in greatest need" — NOT the diagnostic system, and this page says so. https://doi.org/10.1002/wps.20388
  62. Ormel J, Kessler RC, Schoevers R, et al. "More treatment but no less depression: The treatment-prevalence paradox." Clinical Psychology Review 2022;91:102111. doi:10.1016/j.cpr.2021.102111. "Treatments for depression have improved, and their availability has markedly increased since the 1980s. Mysteriously the general population prevalence of depression has not decreased." The authors' own accounting attributes most of the paradox to efficacy overestimation in the published literature and to real-world attenuation of treatment effects — again, not to the diagnostic categories. https://doi.org/10.1016/j.cpr.2021.102111
  63. Garnett MF, Zehner M. "Changes in Suicide Rates in the United States From 2022 to 2023." NCHS Data Brief No. 541, National Center for Health Statistics, 2025. "The overall age-adjusted suicide rate increased from 2003 (10.8 deaths per 100,000 population) to 2018 (14.2) but did not significantly change between 2018 and 2023 (14.1)." Earlier segment from NCHS Data Brief No. 241 (Curtin, Warner & Hedegaard): the age-adjusted rate rose 24% from 1999 (10.5) to 2014 (13.0). We give no pre-1999 baseline because we could not retrieve one from an NCHS source. https://www.cdc.gov/nchs/products/databriefs/db541.htm
  64. Insel T, Lieberman JA. "DSM-5 and RDoC: Shared Interests." NIMH press release, 13 May 2013. Live URL gone; quoted from the Internet Archive capture of 7 June 2013, read 2026-08-27. Co-signed by the NIMH Director and the APA President-elect two weeks after "Transforming Diagnosis" [49]: the DSM and ICD "represents the best information currently available for clinical diagnosis of mental disorders. Patients, families, and insurers can be confident that effective treatments are available and that the DSM is the key resource for delivering the best available care. The National Institute of Mental Health (NIMH) has not changed its position on DSM-5"; "DSM-5 and RDoC represent complementary, not competing, frameworks"; "But this is a long-term undertaking. It will take years." http://web.archive.org/web/20130607222752/http://www.nimh.nih.gov/news/science-news/2013/dsm-5-and-rdoc-shared-interests.shtml
  65. National Institute of Mental Health. "About RDoC." Read 2026-08-27: "RDoC is not meant to serve as a diagnostic guide, nor is it intended to replace current diagnostic systems." https://www.nimh.nih.gov/research/research-funded-by-nimh/rdoc/about-rdoc
  66. Aboraya A, Rankin E, France C, El-Missiry A, John C. "The Reliability of Psychiatric Diagnosis Revisited: The Clinician's Guide to Improve the Reliability of Psychiatric Diagnosis." Psychiatry (Edgmont) 2006;3(1):41–50. PMC2990547, open access, read 2026-08-27. The carrier for the pre-DSM-III agreement studies and for the Ward et al. finding that the three reasons for diagnostic disagreement were "inconstancy of the patient (5%), inconstancy of the clinician (32.5%), and inadequacy of the nomenclature (62.5%)." That 62.5% is the strongest single number in the manual's favour anywhere on this page, which is why it is printed in the section arguing for it. https://pmc.ncbi.nlm.nih.gov/articles/PMC2990547/
  67. Wakefield JC. "The concept of mental disorder: diagnostic implications of the harmful dysfunction analysis." World Psychiatry 2007;6(3):149–156. PMC2174594, open access, full text read 2026-08-27. Source of the deepest available defence: after listing every standard validity test, "whether the distinguished constructs are disorder versus nondisorder goes beyond the test's capabilities... Thus, there is no substitute for the concept of mental disorder as the ultimate standard. None of our empirical approaches work without a warrant in a conceptual analysis of disorder." https://pmc.ncbi.nlm.nih.gov/articles/PMC2174594/
  68. First MB. "Clinical utility: a prerequisite for the adoption of a dimensional approach in DSM." Journal of Abnormal Psychology 2005;114(4):560–564. doi:10.1037/0021-843X.114.4.560. PMID 16351379. Abstract only. "Adopting a dimensional approach would likely complicate medical record keeping, create administrative and clinical barriers between mental disorders and medical conditions, require a massive retraining effort, disrupt research efforts (e.g., meta-analyses), and complicate clinicians' efforts to integrate prior clinical research using DSM categories into clinical practice." Note the precision this page keeps: First argues that clinical utility must be demonstrated before a system is replaced. He does not argue that the DSM's purpose is utility instead of validity. https://doi.org/10.1037/0021-843X.114.4.560
  69. Kendler KS. "The nature of psychiatric disorders." World Psychiatry 2016;15(1):5–12. doi:10.1002/wps.20292. PMC4780286, open access, full text read 2026-08-27. Both halves are quoted on this page: the defence — "a diagnosis is real to the degree that it 'coheres' well with what we already know empirically" — and the concession, "If I were to have a public debate with an arch anti-psychiatrist, I would not want to put myself in the position of defending the reality of every category in the DSM-5 or ICD-10... We have many more reasons to defend the reality of the broad classes of psychiatric illness than the specific categories in our current diagnostic manuals." Also "If we re-run the tape of history over and over again, the DSM and ICD would not likely have the same categories on every iteration." https://doi.org/10.1002/wps.20292
  70. First MB, Rebello TJ, Keeley JW, et al. "Do mental health professionals use diagnostic classifications the way we think they do? A global survey." World Psychiatry 2018;17(2):187–195. doi:10.1002/wps.20525. PMC5980454, open access. 1,764 clinicians in 92 countries. The finding cuts in both directions and both are on this page: clinicians "reported using diagnostic classifications most often for administrative or billing purposes", and both DSM and ICD "received the highest ratings of utility for meeting administrative requirements, assigning a diagnosis, communicating with other health care professionals, and teaching trainees or students, and the lowest ratings for selecting a treatment and assessing probable prognosis." https://doi.org/10.1002/wps.20525
  71. Kotov R, Krueger RF, Watson D. "A paradigm shift in psychiatric classification: the Hierarchical Taxonomy Of Psychopathology (HiTOP)." World Psychiatry 2018;17(1):24–25. doi:10.1002/wps.20478. PMC5775140, open access. The consortium's own status statement: "To date, HiTOP has not been used clinically as a complete system"; "HiTOP is a work in progress." https://doi.org/10.1002/wps.20478
  72. Ruggero CJ, Kotov R, Hopwood CJ, First M, et al. "Integrating the Hierarchical Taxonomy of Psychopathology (HiTOP) into clinical practice." Journal of Consulting and Clinical Psychology 2019;87(12):1069–1084. doi:10.1037/ccp0000452. PMC6859953, open access. Under the heading "How can a clinician using HiTOP be reimbursed?": "Reimbursement is often tied to ICD codes... the appropriate 'unspecified' categories that correspond to the patient's presenting symptoms can be used to meet administrative requirements... This approach has limitations, but can provide a solution until billing and administrative procedures are better aligned with quantitative nosology." And, flatly: "DSM and ICD codes are likely to remain the language of administrative systems for years to come." https://doi.org/10.1037/ccp0000452
  73. World Health Organization. "Classification of Diseases" standards page, and Resolution WHA72.15, 72nd World Health Assembly, 28 May 2019, which "ADOPTS the eleventh revision... (ICD-11), to come into effect on 1 January 2022." The implementation figures quoted on this page are WHO's own, as published May 2024: 132 member states and areas "at various phases of implementing", 72 having "commenced the implementation process", and "14 countries and areas have begun to collect or report data using ICD-11 coding." The United States is not among them; US claims still run on ICD-10-CM [43]. https://www.who.int/standards/classifications/classification-of-diseases
  74. US Food and Drug Administration. PROZAC (fluoxetine hydrochloride) prescribing information, label effective 2026-01-08, retrieved via the openFDA drug label API 2026-08-27 and read in full. Indications quoted verbatim: "Acute and maintenance treatment of Major Depressive Disorder... Acute and maintenance treatment of obsessions and compulsions in patients with Obsessive Compulsive Disorder (OCD)... Acute and maintenance treatment of binge-eating and vomiting behaviors in patients with moderate to severe Bulimia Nervosa... Acute treatment of Panic Disorder, with or without agoraphobia", plus two indications for PROZAC in combination with olanzapine (acute depressive episodes associated with Bipolar I Disorder, and treatment-resistant depression). The same label draws the boundary we print on this page: "PROZAC monotherapy is not indicated for the treatment of depressive episodes associated with Bipolar I Disorder or the treatment of treatment resistant depression." One molecule spans a mood disorder, an obsessive-compulsive disorder, an eating disorder and an anxiety disorder — and the bipolar line is drawn hard. https://api.fda.gov/drug/label.json?search=openfda.brand_name:%22PROZAC%22
  75. US Food and Drug Administration. ZOLOFT (sertraline hydrochloride) prescribing information, label effective 2023-08-08, retrieved via the openFDA drug label API 2026-08-27. This is the current brand label; source 56 is the generic sertraline labelling, which carries the older "though not established in controlled trials" hedge — the two are different label versions, not a contradiction, and both are quoted as written. Dosage and Administration §2.3, verbatim: "Screen for Bipolar Disorder Prior to Starting ZOLOFT — Prior to initiating treatment with ZOLOFT or another antidepressant, screen patients for a personal or family history of bipolar disorder, mania, or hypomania." Warnings §5.4, verbatim: "In patients with bipolar disorder, treating a depressive episode with ZOLOFT or another antidepressant may precipitate a mixed/manic episode." DELIBERATELY NOT QUOTED ON THIS PAGE: the 0.4% mania/hypomania figure in the same section. It comes from trials in which, per the label, "patients with bipolar disorder were generally excluded" — it is an adverse-event rate in a non-bipolar population and reporting it as a switch rate in bipolar depression would be badly misleading. https://api.fda.gov/drug/label.json?search=openfda.brand_name:%22ZOLOFT%22
  76. US Food and Drug Administration. CLOZARIL (clozapine) prescribing information, label effective 2025-06-06, retrieved via the openFDA drug label API 2026-08-27. Indications quoted verbatim: "Treatment of severely ill patients with schizophrenia who fail to respond adequately to standard antipsychotic treatment. Because of the risks of severe neutropenia and of seizure associated with its use, CLOZARIL should be used only in patients who have failed to respond adequately to standard antipsychotic treatment (1.1)" and "Reducing the risk of recurrent suicidal behavior in patients with schizophrenia or schizoaffective disorder who are judged to be at chronic risk for re-experiencing suicidal behavior. (1.2)" The label states the effectiveness finding rests on a six-week randomised double-blind active-controlled trial against chlorpromazine in patients who had already failed other antipsychotics. https://api.fda.gov/drug/label.json?search=openfda.brand_name:%22CLOZARIL%22
  77. Baca-Garcia E, Perez-Rodriguez MM, Basurte-Villamor I, et al. "Diagnostic stability and evolution of bipolar disorder in clinical practice: a prospective cohort study." Acta Psychiatrica Scandinavica 2007;115(6):473–480. doi:10.1111/j.1600-0447.2006.00984.x. PMID 17498159. Same Spanish catchment cohort as [53]; 1,153 adults who received a bipolar diagnosis at least once, across 71,543 assessments. Verbatim: "Prospective and retrospective consistencies were 49% and 38%. Twenty-three per cent of patients received a BD diagnosis during ≥75% of the assessments. There was a high prevalence of misdiagnosis and diagnostic shift from other psychiatric disorders to BD. Temporal consistency was lower than in other studies." That last clause is the authors' own, and it is why this page presents the figure as one cohort in one country rather than as the stability of bipolar disorder in general. https://doi.org/10.1111/j.1600-0447.2006.00984.x
  78. Fusar-Poli P, Cappucciati M, Rutigliano G, Heslin M, Stahl D, Brittenden Z, Caverzasi E, McGuire P, Carpenter WT. "Diagnostic Stability of ICD/DSM First Episode Psychosis Diagnoses: Meta-analysis." Schizophrenia Bulletin 2016;42(6):1395–1406. doi:10.1093/schbul/sbw020. PMID 26980142, PMC5049518, open access. 42 studies, 45 samples, 14,484 first-episode patients, average follow-up 4.5 years. Every figure in the stability table on this page is quoted verbatim from the abstract: "Prospective diagnostic stability ranked: schizophrenia 0.90 (95% CI 0.85-0.95), affective spectrum psychoses 0.84 (95% CI 0.79-0.89), schizoaffective disorder 0.72 (95% CI 0.61-0.73), substance-induced psychotic disorder 0.66 (95% CI 0.51-0.81), delusional disorder 0.59 (95% CI 0.47-0.71), acute and transient psychotic disorder/brief psychotic disorder 0.56 (95% CI 0.62-0.60), psychosis not otherwise specified 0.36 (95% CI 0.27-0.45, schizophreniform disorder 0.29 (95% CI 0.22-0.38)." The abstract as published contains two typographical defects, both reproduced rather than repaired: the interval for acute and transient psychotic disorder cannot bracket its own point estimate, and the psychosis-NOS parenthesis is never closed. The authors' conclusion is also printed on this page: "There is meta-analytical evidence for high prospective diagnostic stability in schizophrenia spectrum and affective spectrum psychoses, with no significant ICD/DSM differences." Retrieved from the NCBI E-utilities API 2026-08-27. This meta-analysis covers first-episode psychosis only; nothing in it licenses a claim about the stability of depression, ADHD, autism or personality disorder, and this page makes none. https://doi.org/10.1093/schbul/sbw020

related on resolv