Multivariate analyses of anthropometric and psychometric variables in probands with affective disorders and their...

Data on 121 subjects, probands with affective disorders, their spouses and first degree relatives, are analyzed by multivariate methods to determine the relationships between physical types and propensity to illness. 56 variables are used: 39 anthropometric measures, age, sex, and 15 psychometric scales. In a canonical analysis between the anthropometric measures and the psychiatric scales, each canonical variable is found to be largely identified with a single psychometric component, as found in a principal components analysis of the psychometric scales. The two major anthropometric components, size and linearity, do not have any clear relationship with the psychometric components. However, a discriminant analysis that takes each individual as being in one of four clinical groups, normal, unipolar depressed, bipolar affective disorder or other, indicates a clear relationship between the anthropometric measures and mental illness; wide face and deep chest are associated with patients who have bipolar affective disorder. Half of the variables studied are sufficient to give virtually the same amount of discrimination as all 56 variables.