Abstract

The accompanying paper presented in this issue of the Southern Medical Journal reinforces concepts of data analysis that are of great importance for clinical psychiatrists. Despite enormous advances in the biologic understanding of mental illness, psychiatry remains an empirical medical specialty, whose outcomes data often cannot be interpreted in light of conceptual models of pathophysiology. The absence of specific biologic measures for clinical outcomes has forced the field to create empirically derived assessments with varying levels of clinical relevance. The advent of the DSM-III (Diagnostic and Statistical Manual) in 1980 heralded the era of empiricism in psychiatry, whereby disease states were (ideally) defined by common phenomenology, and not on psychoanalytically based conceptualizations for psychosis, mood disorders, or anxiety. DSM-III was supposed to use an atheoretical approach to diagnosis, but even such approaches require a theoretical model of disease in which the natural continua of psychopathology are divided into normal/abnormal by the simple act of defining a disease as 5 of 9 symptoms (eg, for major depression) instead of 4 of 9. Nonetheless, DSM was a boon to psychiatric research by establishing a common vocabulary for disorders, and thereby furthering the refinement of instruments to assess treatment outcomes.