Background: Diagnostic comorbidity is prevalent in psychiatry and may be inadequately captured by the DSM-III/III-R nosology.
Methods: The lifetime presence of II psychiatric diagnoses was determined by structured personal interviews of a population-based sample of 1898 female twins. We used latent class analysis to derive an empirical typology.
Results: Six classes provided the best fit to the data. Their mnemonics were: minimal disorder (60% of the sample), major depression-generalised anxiety disorder (19%), alcohol-nicotine (7%), highly comorbid major depression (5%) and eating disorders (3%). The validity of this typology was strongly supported by demographic, health, personality and attitudinal validators along with the significant monozygotic twin concordance for class membership. The typology superficially resembled DSM-III-R, but contained many differences. Major depression appeared in three forms (alone, with generalised anxiety disorder and with considerable comorbidity). Alcoholism-nicotine dependence and the various anxiety disorders formed discrete classes, but were also prominent in other classes. Bulimia and anorexia were exceptional in their appearance in a single class.
Conclusions: The DSM-III-R and closely related DSM-IV nosology did not capture the natural tendency of these disorders to co-occur. Fundamental assumptions of the dominant diagnostic schemata may be incorrect.