M195. LATENT STRUCTURE OF NEGATIVE SYMPTOMS IN EARLY PSYCHOSIS AND CLINICAL HIGH-RISK FOR PSYCHOSIS

Abstract Background Negative symptoms are prevalent and predictive of clinical and functional outcomes across different phases of psychotic disorders. Yet, heterogeneity in conceptualizing the latent structure of negative symptoms presents hindrances to the development of effective interventions. While a 2-dimensional construct of negative symptoms (i.e., Motivation and pleasure [MAP] and Emotional expressivity [EXP] dimensions) have previously been derived from exploratory factor analyses and adopted widely in research, conflicting findings in favor of a 5-factor structure have emerged from confirmatory factor analyses recently. Further evidence is needed to evaluate whether this conclusion can be generalized to the prodromal and early phases of psychosis. Methods Data were drawn from 3 studies that administered the Brief Negative Symptom Scale (BNSS), a second-generation clinical rating instrument, to assess negative symptoms in Chinese patients with early psychosis or clinical high-risk for psychosis (CHR) in Hong Kong. The early psychosis sample comprised 131 and 246 outpatients recruited in 2 separate studies who received treatment within 5 years since service entry for first-episode psychosis, whereas the CHR sample included 110 help-seeking individuals ascertained using CAARMS criteria. Confirmatory factor analyses (CFAs) were employed to examine competing hypotheses about the factor structure of negative symptoms as measured by BNSS. The fit of five competing models were evaluated, including 1) a unifactorial model, 2) a 2-factor model with EXP and MAP factors, 3) a 3-factor model proposed by Garcia-Portilla et al. (anhedonia and asociality, avolition and blunted affect, and alogia), 4) a 5-factor model specifying the five NIMH consensus development conference domains (blunted affect, alogia, anhedonia, avolition, asociality), and 5) a hierarchical model with two second order-factors reflecting EXP and MAP, and five first-order factors reflecting the five consensus domains. Results In the early psychosis sample, the 1-, 2-, and 3-factor models provided poor fit for the data. The 5-factor and hierarchical models were excellent fit, with the hierarchical model being slightly more parsimonious. Similarly, CFA of the CHR sample demonstrated that the 1-, 2-, and 3-factor models were poor fit for the BNSS, whereas the 5-factor and hierarchical models provided strong fit. The 5-factor model is slightly favored over the hierarchical model in the CHR sample