Context sensitivity and the DSM
"The genetics of psychological suffering is the genetics of being human, and in particular our capacity to react with feeling to the environment." — Jim van Os
What is DSM?
The DSM stands for Diagnostic and Statistical Manual of Mental Disorders. It is a classification system used in psychiatry to describe mental disorders.
See also the Wikipedia article: Diagnostic and Statistical Manual of Mental Disorders.
Origin
The DSM finds its origin in the American military and statistics. There was a need for reliable and uniform diagnostics to screen and treat soldiers.
Descriptive, not explanatory
The DSM is a descriptive model:
- it notes what someone experiences (symptoms),
- but does not explain why someone has those complaints.
A "diagnosis" like appendicitis is explanatory and clear. A DSM classification, on the other hand, is a tentative classification of symptoms.
Case
A patient presents with:
- two weeks of persistent low mood
- fatigue
- loss of interest in activities
- sleep problems
- feelings of worthlessness
According to the DSM-5, this meets the criteria for a depressive disorder (DSM-5 criteria, NCBI).
The classification describes the symptoms, but says nothing about the cause.
The underlying reason can be very different:
- loss of work
- relationship problems
- physical illness
- context-blindness or difficulty with complex thinking
The label depression is therefore a description, not an explanation.
The reliability of DSM classifications is variable and method-dependent. The DSM-5 field trials show variable kappa values per diagnosis. Chmielewski et al. (2015) demonstrate that when two independent clinicians conduct separate interviews, agreement is significantly lower than when one clinician rates the same recording twice. The outcome therefore depends strongly on how reliability is measured — not only on the criteria themselves.
For clinical practice, this means: a DSM label is often less robust than the tone of criteria lists suggests.
Note
This is not about "poor criteria" but about a structural methodological point. The DSM was designed to increase agreement — but that agreement remains variable in practice and depends on the measurement method.
The consultation as label production
A criteria-driven conversation can end with a label — but without an answer to the questions the patient is actually asking: what has changed, what is maintaining the problem, what is the risk, and what do we agree on for the coming weeks?
When DSM criteria become the structure of the consultation, the conversation shifts from "what is happening here and what helps" to "which diagnosis is it". The label then becomes the goal, while the person's situation becomes backdrop.
A functional working hypothesis — based on timeline, context, and modifiable links — often guides clinical action better than a classification alone.
Note
This is not an argument against diagnoses. A DSM classification can be useful as a summary language, for communication with other care providers, or for administrative purposes. The pitfall is that the label replaces the plan rather than informing it.
Purpose of the DSM
According to psychiatrist Jim van Os:
- There was a proliferation of diagnoses in psychiatry.
- Clear criteria were needed to arrive at a diagnosis.
- This was also necessary to enable validated reimbursement by health insurance.
What is it not?
The DSM is not scientific evidence that these entities truly exist. Sometimes there are attempts to "prove" this afterwards with brain research or biomarkers, but that is a form of reverse engineering. After 50 years of research, no conclusive evidence has been found that the DSM classifications are real diseases like in somatic medicine.
Boundaries shift between editions
The DSM is not a fixed measure. Fabiano & Haslam (2020) analysed 123 studies in which the same sample was diagnosed simultaneously using two consecutive DSM editions. The thresholds and definitions change between editions, but not all in the same direction. Some diagnoses become easier to make, others become harder.
Well-known examples of diagnostic inflation are ADHD, autism, and eating disorders. For other diagnoses, the criteria have actually become stricter.
For practice, this means: someone who meets the criteria for a label today may no longer meet them in the next edition — and vice versa. A label is not a timeless, biologically defined disease entity.
What does genetics say?
The only field that provided insights is genetics. It shows that every person has thousands of genetic variants that are involved in psychological suffering. These variants occur in autism, depression, borderline, schizophrenia, bipolar disorder, ... and are largely overlapping (about 70%). Conclusion: the genetics of psychological suffering is in fact the genetics of being human, and more specifically our capacity to react with feeling to the environment.
Spectrum thinking
Terms like autism, schizophrenia, depression, borderline are not diseases in themselves, but descriptions. Therefore, people are increasingly talking about a spectrum or "characteristics of". This is also evident in personality disorders: a person can at a certain moment meet the criteria for borderline, later for narcissism, and then not again.
But even "spectrum" still suggests a single line. The reality is multidimensional: every profile is a unique combination of stronger and weaker contextual domains. See The multidimensional profile.
More treatment, not less depression
Despite a strongly increased treatment supply since the 1980s, the prevalence of depression in the population has not declined. Ormel et al. (2022) call this the "treatment-prevalence paradox" and describe seven possible explanations. The most supported: the efficacy of treatments is overestimated in the literature, and the effect in clinical practice falls below what is found in trials.
This supports the argument that a purely disease-oriented approach has its limits. Living conditions, social context, and the question of what maintains the suffering weigh more heavily than the label suggests.
Interests in the development
The DSM is compiled by committees. This has practical consequences for clinical practice: it influences which complaints more quickly receive a label, how guidelines are written, and how records are used in recognition and insurance processes.
Cosgrove & Krimsky (2012) reported that 69% of DSM-5 task force members disclosed financial ties with the pharmaceutical industry, compared to 57% for DSM-IV. Davis et al. (2024, BMJ) analysed DSM-5-TR panel members via the Open Payments database: 60% of the 92 physicians examined received payments, totalling 14.2 million dollars over the study period.
These are facts, not conspiracy theories. They do not claim that every diagnosis is wrong because interests are involved. They do justify extra vigilance: when a classification carries so much weight in policy, reimbursement, and clinical practice, the standards for independence must be high.
Note
These data come from PubMed-verified publications: Cosgrove & Krimsky (PubMed 22427747, DOI 10.1371/journal.pmed.1001190) and Davis et al. (PubMed 38199616, DOI 10.1136/bmj-2023-076902).
Listen further
Podcast by Jim van Os about the DSM and psychological classification: Listen on Spotify