The stability of diagnosis with an application to bipolar II disorder

Abstract

A temporal stability study (where independent diagnostic interviews are conducted at widely separated time points) is discussed and compared to a test/retest reliability study. We introduce a new battery of statistics, based on the sensitivity, specificity, and true base rate of a disorder, that quantifies either reliability or stability over time, and provide a table that relates these underlying parameters to the opening characteristics of the battery. We analyze data on 50 relatives who participated in the family study component of the National Institute of Mental Health's collaborative "Psychobiology Depression Program." The subjects received an interview 5 years after their initial evaluation by raters without knowledge of the initial assessment. The stability of mania, hypomania, major depression, and alcoholism is considered. Although the kappa coefficient for hypomania was small (0.09), all diagnoses of hypomania in the relatives occurred in the families of bipolar probands, suggesting that the low value of kappa is due to a low sensitivity rather than to a low specificity. This is compatible with earlier findings and demonstrates the value of multiple independent assessments when studying this disorder.