Abstract
The delineation of schizophrenia-related symptomatology is critical to informative clinical phenotyping in linkage studies. A minority of first-degree relatives of schizophrenia and schizoaffective probands (RelSZSA) qualifies for a clinical diagnosis in the schizophrenia spectrum. Schizotypal personality disorder (SPD) is a key component of this spectrum, largely because of its relatively specific familial aggregation in relatives. The criteria for SPD were not developed for the purpose of identifying RelSZSA, however, and SPD is not a homogeneous clinical disorder, potentially introducing false positives and false negatives into affectedness classifications. In this study we used logistic regression (LR) to identify the combination of clinical signs and symptoms that maximized the discrimination between nonpsychotic first-degree RelSZSA (n=241) and controls (n=161). Three variables contributed significantly to optimizing this distinction: no close friends or confidants other than family members, social isolation and irritability. The combination of deviant LR scores and schizophrenia-spectrum psychotic disorders had greater sensitivity for identifying RelSZSA, 23.7%, than SPD and schizophrenia-spectrum psychotic disorders, 16%. Importantly, the diagnosis of SPD and deviant LR scores were not significantly correlated. Most individuals with deviant LR scores did not meet criteria for a diagnosis of SPD and only a minority of those who were diagnosed with SPD had deviant LR scores. Since misclassification of gene carriers as non-gene carriers in linkage analyses increases the risk of false negatives, it may be advantageous to tailor the definition of the clinical phenotype to those aspects of social-interpersonal dysfunction that optimize the discrimination of RelSZSA from controls.