Psychiatrists presented with identical case descriptions agree on a diagnosis only a little over half the time, according to a large international study. The findings highlight significant shortcomings in how mental disorders are classified and diagnosed in everyday clinical practice.
The research was conducted by scientists at the University of Copenhagen together with clinicians from multiple countries. A total of 1,038 doctors working in adult psychiatry across 19 European and South American countries participated, producing 1,902 individual assessments.
Each participant received two randomly selected cases from nine prepared clinical scenarios and was asked to choose one diagnosis from a list of 30 psychiatric conditions. Some cases were created specifically for the project, while two were adapted from World Health Organization ICD-10 training materials.
Before the main study, a group of experienced psychiatrists identified which diagnoses, in their view, best matched each case. These reference diagnoses were not treated as absolute truth but served as a benchmark for assessing how much agreement could realistically be expected.
Low agreement on the same patients
The key result was concerning. The statistical measure of agreement between doctors, known as kappa, was 0.48, which was considered weak on the scale used in the study. In practical terms, two psychiatrists reviewing the same case description selected the same diagnosis only about 55 percent of the time.
Discrepancies were particularly pronounced for schizophrenia and related schizotypal disorders. In one vignette, 37 percent of doctors diagnosed the patient with schizophrenia, while 41 percent selected obsessive-compulsive disorder instead, despite reading identical clinical information.
In another scenario, 43 percent of participants chose schizophrenia and 45 percent selected a personality disorder. Across several cases involving schizophrenia or schizotypal presentations, diagnostic agreement fell below 50 percent, highlighting how blurred the boundaries between categories can become in practice.
The researchers stress that this variation did not appear to reflect random guesswork. Certain alternative diagnoses repeatedly emerged, suggesting that clinicians systematically placed greater emphasis on different aspects of the same description, such as obsessive symptoms versus psychotic signs.
Experience does not fully resolve gaps
Professional experience did little to eliminate the disagreement. The kappa value among fully trained psychiatrists was 0.50, compared with 0.45 among psychiatry residents and other doctors. Statistically, the differences were too small to conclude that one group clearly outperformed the others.
Similarly, the number of years spent in clinical practice was not clearly associated with greater diagnostic consistency. This suggests that the problem may lie primarily within the diagnostic framework itself rather than simply reflecting insufficient training or clinical exposure.
Not all conditions were equally difficult to identify. Agreement with the reference diagnosis reached 94 percent for bipolar disorder, 92 percent for obsessive-compulsive disorder and 88 percent for depression, showing that clearer symptom patterns can still produce strong consensus.
According to the authors, this contrast highlights the enormous complexity of current psychiatric classification. Systems such as ICD-10 contain more than 200 diagnostic categories, many involving overlapping or subtly differentiated symptom clusters that can be difficult to distinguish in real-world cases.
Consequences for patients and research
The findings suggest that more precise and potentially fewer diagnostic categories may be needed to improve reliability. Sharper descriptions of core symptoms and clearer diagnostic thresholds could help clinicians arrive at the same diagnosis when evaluating similar patients.
For patients, inconsistent diagnoses can lead to different treatment plans, access to services and prognoses for essentially the same clinical presentation. Differences in diagnostic labels may also affect stigma and self-understanding because psychiatric diagnoses can carry substantial personal and social significance.
The implications extend to mental health research and drug development. If participants in clinical trials are misclassified at the outset, determining which therapies genuinely work for particular conditions becomes more difficult.
When diagnostic boundaries are porous, a medication intended for one disorder may effectively be tested in a mixed group of patients experiencing partially different problems. This can weaken study findings and slow progress toward more effective and targeted treatments.
The Copenhagen-led team argues that future revisions of psychiatric classification systems should place greater emphasis on reliability without sacrificing clinical nuance. Balancing scientific precision with the complexity of human psychology remains one of the central challenges facing modern psychiatry.
