Searching for diagnoses and subgroups: a suggestion for criteria.

New subgroups of psychiatric disorders are often claimed. In contrast, classification systems have repeatedly had to abandon established subgroups such as paranoid vs. disorganised and catatonic schizophrenia due to lack of empirical evidence. Four criteria are proposed that should be met to claim valid subgroups: 1. distinct distribution of the defining characteristic between groups; 2. significant differences in variables other than those defining the subgroups cross-sectionally and longitudinally; 3. long-term stability; 4. significant differences between groups in aetiology, pathophysiology, and evidence-based therapy. In contrast to examples from somatic medicine, such as type 1 and type 2 diabetes, few psychiatric disorders meet these requirements.