The DSM: What a Diagnosis Can Tell You

Where ordinary traits become disabling

Everyone gets distracted or worried. Diagnosis asks about a persistent pattern: how it developed, how severe it is and what it prevents someone from doing. A familiar symptom is only one part of that account.

A cutoff creates a difficult boundary. Two people just above and below it may resemble each other more than two people who both qualify. Yet a shared label helps clinicians communicate, compare treatments and arrange support. Its usefulness and the uncertainty around its boundary can coexist.

Biological evidence also comes in forms that aren't diagnostic tests. A review of ADHD twin studies reported an average heritability estimate of 74 percent. Heritability describes variation in a population under a model; it isn't the percentage of an individual's condition caused by genes, or a DNA result that confirms ADHD. Faraone and Larsson, 2019

A symptom with several possible causes

Psychosis includes hallucinations, such as hearing voices others don't hear, and delusions, beliefs held despite compelling contrary evidence. It is a set of symptoms with several possible causes. APA

The last three sketches concern the timing and pattern of psychosis. Medical and substance-related causes also need consideration; the DSM lists those diagnoses explicitly. A first assessment may remain provisional while the history and course become clearer. APA: psychotic disorders

A much older example is general paresis, a progressive illness involving psychiatric symptoms and cognitive decline. In 1913, Noguchi and Moore demonstrated syphilis bacteria in affected brains. Penicillin treatment for syphilis was first reported in 1943, three decades later. Finding a cause and having an effective treatment were separate events. Noguchi and Moore, 1913 Mahoney et al., 1943

These examples support a practical requirement: investigate plausible medical causes of psychiatric symptoms. They don't show that all psychosis has a hidden infection or that routine symptoms demand every rare-disease test. The workup depends on the history and examination. Graus et al. CDC: neurosyphilis

What a normal scan doesn't settle

Pain, disturbed sleep, compulsive rituals and changes in eating can be disabling even when a scan doesn't explain them. These diagnoses concern different patterns, not a common rule that the symptoms must lack a physical cause.

Anorexia carries a high mortality risk. A 2011 meta-analysis reported mortality about 5.9 times the rate expected in its comparison populations. It compared eating disorders, so it cannot establish that anorexia is the deadliest of all psychiatric diagnoses. A separate opioid review reported an even higher relative mortality estimate, in different populations and settings. Those studies establish serious risks, not a universal ranking. Arcelus et al., 2011 Degenhardt et al., 2011

Nor is anorexia assessed only by talking. Weight, vital signs and laboratory results matter for evaluating medical complications, alongside the behavior and experience that define the disorder. A lack of a single diagnostic biomarker says little about how dangerous the condition is. APA

A pattern over time

Personality-disorder diagnoses concern enduring patterns of thinking, feeling and relating that cause distress or impair functioning. Their assessment includes cultural context. These labels should not be inferred from a few unpleasant encounters. APA

The personality sketches above summarize broad patterns, not entire people. Their resemblance to ordinary insults makes casual diagnosis especially unhelpful: calling someone a narcissist does not establish narcissistic personality disorder. A clinical account needs persistence, context and consequences. APA

Biology, behavior and circumstances

A substance use disorder describes impaired control and continued use despite harm. Brain changes, access to a substance, learned habits and a person's circumstances can all matter. Calling it a disorder does not mean recovery is impossible. APA

DSM-5 combined the older categories of substance abuse and dependence in 2013, and moved gambling into its addiction chapter. Internet gaming disorder remains a condition for further study in the DSM. These are classification decisions informed by evidence, with consequences for what clinicians diagnose and researchers study. APA: DSM-5 changes APA: internet gaming

A urine drug test can provide evidence of recent exposure. It cannot by itself establish a substance use disorder; screening tests can also need confirmation. Conversely, tolerance or withdrawal during supervised medical treatment need not mean addiction. MedlinePlus: drug testing APA: opioids

The Vietnam-veteran study gives a useful limit on predictions about relapse. In a weighted sample of Army enlisted men returning in September 1971, about one in five reported heroin addiction in Vietnam. Among those addicted there, 12 percent reported a return to addiction during the next three years. Some resumed heroin use without meeting the study's definition of addiction. Robins et al., original report reprinted in 2010

This was an observational follow-up of a particular cohort, using the investigators' definitions of addiction. Availability and circumstances changed, and some men received treatment. It does not isolate one cause of remission or predict outcomes for someone facing today's opioid supply. It does show that persistent addiction was not inevitable in that sample.

Agreement, evidence and judgment

Two clinicians reaching the same diagnosis is reliability. Whether the category captures a useful and well-supported distinction is a question of validity. A reproducible label can still group together people with different causes, courses or treatment needs.

Major depression's pooled adult estimate was 0.28, in the investigators' "questionable" range. That is a reason to examine assessment quality and diagnostic boundaries. It is not a coin-toss accuracy score, a count of false diagnoses or a universal estimate for every clinical setting.

The design matters. The interviews used ordinary clinical methods on separate visits, allowing differences in what patients reported and clinicians elicited. The sample included people with overlapping conditions. Low agreement can reflect several problems, including missed co-occurring diagnoses; it does not identify which clinician, if either, was wrong. Clarke et al., 2013 Regier et al.

The manual's history also exposes errors in what medicine chooses to classify. In 1973 the APA board removed homosexuality itself as a disorder; members upheld that decision in a 1974 referendum. Research on people outside psychiatric treatment, gay activism and reconsideration of what counts as impairment all contributed. The later category "ego-dystonic homosexuality" was removed in 1987. Drescher, 2015

That history shows how prejudice can enter a diagnostic system and how evidence and public challenge can correct it. It does not show that every diagnosis is arbitrary or that the change happened without research. The relevant distinction was between a sexual orientation and an illness, not between two laboratory results.

A more recent change concerns bereavement. DSM-5 removed the earlier exclusion that discouraged diagnosing depression soon after a loved one's death. It also directs clinicians to distinguish ordinary grief from a depressive episode. Two weeks of sadness after a loss is not, by itself, major depression. The change makes careful judgment more important: a clinician must recognize depression without treating ordinary grief as an illness. APA: bereavement exclusion

Researchers need not accept the DSM's boundaries as final. NIMH's Research Domain Criteria program studies dimensions such as cognition and emotion across current diagnostic categories. It is a research framework, not a replacement diagnostic manual for clinical care. NIMH: RDoC

Historical source for the pellagra caption: NIH history

A diagnosis can guide treatment and support without explaining every symptom or fixing a person's identity. When assessing its usefulness, ask what observations support it, what competing explanations were considered and what would lead the clinician to revise it. Those are questions a symptom checklist on a screen cannot answer.

The remaining uncertainty is often about the category itself: whether it separates conditions with meaningfully different courses or merely collects similar symptoms. A future biological test may help answer that question. Its value will depend on what it predicts for the people who take it.

What the evidence covers

The clinical links describe conditions; the field trials measure agreement in particular samples. Neither is a test of an individual reader. The genetic and medical examples illustrate different forms of evidence, including their limits.