The impact of diagnostic misclassification on the pattern of familial aggregation and coaggregation of psychiatric illness.

Family studies in psychiatry have traditionally been interpreted as if psychiatric diagnoses were made without error. This is unlikely to be true. The first section of this paper presents a model, based on several simplifying assumptions, which examines the impact of diagnostic misclassification on patterns of familial aggregation and coaggregation of two disorders. In the second section, the model is illustrated by examining the effect of varying misclassification rates on the patterns of aggregation and coaggregation in four simulated family studies of psychotic illness. Misclassification is considered which is equal or unequal for the two disorders, which is the same in probands and relatives or greater in relatives and which occurs between disorders of similar or quite different population risk. Misclassification in the range which commonly occurs in psychiatry can produce substantial effects on the observed pattern of familial aggregation and coaggregation. Furthermore, the proportion of diagnosed cases which are misclassified varies widely depending upon whether an affected individual is from the general population or is a relative of an individual affected with the same or a different psychiatric disorder. The third section of the paper illustrates a method for examining the fit of the proposed model to observed data and "correcting" for the effects of misclassification in family studies. For example, the familial coaggregation between bipolar illness and schizophrenia in a recent large family study can be entirely explained by the observed rates of diagnostic misclassification between the two disorders. The results of the proposed model strongly suggest that diagnostic misclassification should be considered in the interpretation of family studies of psychiatric illness.