30 2. Phenomenology patients to retrospectively identify prodromal symptoms, and sug- gest that they become better at identifying them with increased expo- sure to manic episodes. While these symptoms are likely to have meaning after an initial manic episode, it remains unclear how dis- tinct they are from the normal ups and downs of common human experience to be efficient markers of an at risk mental state. Rather it seems more pertinent to study these initial symptoms in a broader biopsychosocial context. As a next step then, research is needed to more thoroughly describe the development of mania, in terms of both prodromal symptoms and the presence of other risk factors. It seems unlikely that defining an early treatable syndrome and potentially developing preventive interventions will be possible without inte- grating such a framework. COMPARATIVE VALIDITY OF ALTERNATIVE SYSTEMS FOR INTERPRETATION OF PANSS RATED SYMPTOMS OF SCHIZOPHRENIA L. White,* L. A. Opler, R D. Harvey, J. I. Friedman Department of Psychiatry, Mt. Sinai School of Medicine, New York, NY, USA The Positive and Negative Syndrome Scale (PANSS) is widely used to assess severity of symptoms of schizophrenia but alterna- tive systems have been proposed for grouping the symptoms into syndromes. This study examined the convergence and predictive validity of four alternative systems for grouping symptoms as rat- ed on the PANSS. 1-Thirty individual items; 2-A priori three syn- dromes (Positive, Negative, General Psychopathology) based apon 30 items as originally proposed by the authors of the PANSS; 3- Five BPRS derived factors based upon 18 items (Thought Disor- der, Anergia, Activation, Paranoid Belligerence, Depression); 4- Pentagonal Model five factors based upon 25 items (Positive, Negative, Activation, Dysphoric Mood, Autistic Preoccupation) (White Let al. 1997). Three hundred thirty three elderly (mean age= 76.1 years +7.0) inpatients in a large state hospital with life- time research diagnosis of schizophrenia were rated on the PANSS. Those remaining in the hospital were rated in the second year and followed for an additional year. Correlational analyses reveals a very high degree of correlation between alternative Pos- itive symptoms scales (r=.80 to .70), and alternative Negative symptom scales (r=.90 to .83) and alternative Mood scales (r=.96). Mood is well differentiated from from Negative symptoms (r=- .30 to -.25).The Autistic Preoccupation factor of the Pentagonal model is moderately correlated with Negative Symptoms (r=.65 to .52). Within the BPRS derived factors there is a modest correla- tion between Activation and Paranoid Belligerence (r=.42) and both are minimally correlated with Thought Disorder (r=.21). Pen- tagonal model Activation is most highly correlated with BPRS based Paranoid Belligerence (r=.85) and is independent of Pen- tagonal model Positive symptoms (r=.00). Predictive validity of alternative systems were examined by MANOVA for first and sec- ond year cohorts with outcome (discharged vs retained) as the classification variable. Findings for each assessment period were specific to the symptom model, The results suggest the predictive validity of the Pentagonal model that sharply differentiates between Activation and Positive symptoms. The five factor Pen- tagonal model of PANSS rated symptoms of schizophrenia may provide a valid system for assessment of the multidimensional fea- tures of schizophrenia but additional validity studies are neces- sary. A LATENT FACTOR ANALYSIS OF THE DEFICIT SYNDROME CRITERIA S. A. Yale,* R. Goetz, L. Marcinko, X. Amador, D. Malaspina Psychiatry, Columbia University, College of Physicians & Surgeons, New York, NY, USA Efforts at defining more homogeneous subgroups of patients with schizophrenia have focused on finding symptom clusters that can better classify groups of patients. Prominent among these efforts is the Deficit Syndrome (DS) classification: The DS is considered to be present in patients who have at least two of six negative symptoms: restricted affect, diminished emotional range, poverty of speech, curbing of interests, diminished sense of purpose, and diminished social drive. The symptoms also must be both primary (i.e. not caused by other symptoms of the illness) and enduring during periods of remission. Based on our work relating DS symptoms with course and outcome data, we wondered if the DS could be comprised of separate factors in schizophrenia patients. We catergorized the DS in 166 DSM-IV schizophrenia patients using the Schedule for the Deficit Syndrome (SDS). We hypothesized that the SDS may contain separable dimensions, respectively indexing decreased emotional reactivity (restricted affect, anhedonia) and impaired social func- tioning (decreased social drive, diminished sense of purpose and curbing of interests). We performed a factor analysis using principal components analysis with varimax rotation on the severity ratings of the six symptoms assessed by the SDS. Analyses were done on the whole sample (n=166) and only those patients who met criteria for the deficit syndrome (n=56). The first analysis of all subjects revealed only one significant factor comprising the symptoms of curbing of interests, diminished sense of purpose and diminished social drive that predicted the DS. The second analysis of DS patients only revealed two factors: factor 1 (restricted affect and anhedonia)and factor 2 (curbing of interests, diminished sense of purpose and dimin- ished social drive). These results confirmed our hypothesis that there are separate factors within the SDS. The data suggests that flat affect and poverty of speech may have a different pathophysiological basis than the other deficit symptom items. Furthermore, DS patients inability to detect the affect of others may be related to some neu- ropsychological deficit that impairs not only their ability to infer social cues from others but limit their own responses to these social cues. MAPPING THE ONSET OF PSYCHOSIS: THE COMPREHENSIVE ASSESSMENT OF AT RISK MENTAL STATES (CAARMS) A. R. Yung,* H. P. Yuen, L. J. Phillips, S. Francey, P. D. McGorry Psychiatry, ORYGEN Research Centre, University of Melbourne, Parkville, EIC, Australia The Comprehensive Assessment of At Risk Mental States (CAARMS) is a semi-structured interview schedule developed specifically to prospectively measure subtle pre-psychotic phenom- ena in individuals thought to be in the prodromal phase of a first psy- chotic episode, that is at high risk for psychosis onset. The aim of this study was to assess the reliability and predictive validity of the CAARMS in subjects considered to be at ultra high risk (UHR) for imminent onset of psychosis. UHR criteria were age between 14 and 30, presence of attenuated psychotic symptoms, brief spontaneous- ly resolving psychotic symptoms or a combination of trait and state risk factors for psychosis. A sample of 49 UHR subjects was fop International Congress on Schizophrenia Research 2003