Psychiatrists Agree on Mental Health Diagnosis in Only 55% of Cases

When adult psychiatrists assess the same patient using standardized criteria, they agree on a diagnosis in only about 55% of cases, according to a peer-reviewed study published on September 9, 2026, in Frontiers in Psychiatry. Led by researchers at the University of Copenhagen, the investigation surveyed 1,038 psychiatrists and physicians working in adult psychiatry across 19 countries in Europe and South America, raising new questions about the reliability of mental health diagnoses and psychiatric research.

To test diagnostic consistency, researchers presented participants with nine written patient case descriptions, assigning two cases to each clinician. The doctors were asked to determine the most appropriate diagnosis using the World Health Organization’s ICD-10 diagnostic system, which contains more than 200 diagnoses.

The overall rate of agreement stood at 55%. According to Professor and Consultant Psychiatrist Julie Nordgaard, that figure closely mirrors the levels of uncertainty documented in the 1970s, which originally spurred the creation of standardized diagnostic criteria by organizations like the American Psychiatric Association and the WHO.

Our study shows that when two psychiatrists diagnose the same patient, they will agree only 55% of the time. We consider that worryingly low. In fact, it is similar to the levels reported in some of the studies conducted in the 1970s, which prompted the development of standardised diagnostic criteria in the first place.

Professor and Consultant Psychiatrist Julie Nordgaard

Why Psychiatric Diagnoses Vary Across Conditions

Diagnostic agreement varied significantly depending on the specific mental disorder. While clinicians showed high levels of diagnostic accuracy on certain conditions—reaching 94% for bipolar disorder and 92% for obsessive-compulsive disorder—diagnoses within the schizophrenia spectrum proved much more difficult to standardize.

Cases that could be classified as either schizophrenia or schizotypal disorder generated especially high levels of disagreement, with participant consensus frequently dropping well below 50%. According to PhD candidate and first author Mateo Boberg, overlapping symptoms are a primary driver of this variability.

Cases that could be diagnosed either as schizophrenia or schizotypal disorder generated especially high levels of disagreement. In many of these cases, agreement among participants was well below 50%.

PhD candidate Mateo Boberg

Boberg noted that symptoms like obsessive thoughts frequently appear in multiple disorders, including both obsessive-compulsive disorder and schizophrenia. He emphasized that these discrepancies are not random errors, but rather point to underlying ambiguities in how mental disorders are currently categorized. The data also showed no significant difference in agreement between more experienced and less experienced clinicians.

Implications for Psychiatric Research and Patient Care

Beyond the immediate impact on patients, who risk receiving conflicting diagnoses and shifting treatment plans, the researchers warned that diagnostic uncertainty threatens the integrity of psychiatric studies. Professor Mads Gram Henriksen explained that research trials require precise definitions of what is being studied to yield interpretable results.

In research, it is essential that we know exactly what we are studying. If a considerable part of participants enrolled in a study on treatment of personality disorders actually suffer from schizophrenia, the study’s results become difficult to interpret. Which condition is the treatment having—or not having—an effect on? Personality disorders or schizophrenia?

Professor Mads Gram Henriksen

To address these challenges, the University of Copenhagen research team suggests that the psychiatric field may need to take a step back. Nordgaard noted that the complexity of the ICD-10 system contributes to uncertainty, pointing toward a need for more precise clinical descriptions and potentially a reduction in the total number of diagnostic categories to improve patient outcomes.