We determined the rates of agreement between diagnoses, using the Diagnostic Interview for Genetic Studies (DIGS) and diagnoses arrived at, using additional sources of information, to establish whether there are differences in agreement between direct interview diagnoses at US and non-US sites in comparison best estimate consensus process and to identify diagnoses that could increase diagnostic error when only the DIGS is used. DIGS diagnoses were compared with consensus diagnoses that used the same DIGS interview, plus Family Interview for Genetic Studies (FIGS) and review of medical records in 342 psychotic subjects. We found similar numbers of subjects diagnosed with schizophrenia (225 by direct interview, and 232 by consensus process). The majority of those "misdiagnosed" by direct interview had mood disorder by the consensus. Over 10% of the total subjects diagnosed by direct interview as not meeting criteria for schizophrenia had schizophrenia by consensus. There were no statistically significant differences between countries (US vs. non-US sites) in the agreement rate between direct interview diagnosis and consensus diagnosis. In conclusion, a final best-estimate process is essential to make diagnostic distinctions and to reduce diagnostic misclassifications for both research studies and in clinical practice.
