Clinical global impression of cognition in schizophrenia (CGI-CogS): Reliability and validity of a co-primary measure of cognition

Background: Cognitive deficits are core features of schizophrenia that have been associated reliably with functional outcomes and now are a focus of treatment research. New rating scales are needed to complement current psychometric testing procedures, both to enable wider clinical use, and to serve as endpoints in clinical trials. Methods: Subjects were 35 schizophhrenia patient-and-caregiver pairs recruited from the UCLA and West Los Angeles VA Outpatient Psychiatry Departments. Participants were assessed with the Clinical Global Impression of Cognition in Schizophrenia (CGI-CogS), an interview-based rating scale of cognitive functioning, on 3 occasions (baseline, 1 month, and 3 months). A computerized neurocognitive battery (Cogtest), an assessment of functioning, and symptom measures were administered at two occasions (baseline and one month). Results: The CGI-CogS ratings generally showed a high level of internal consistency (Cronbach's alpha = .69 to .96), adequate levels of inter-rater reliability (ICC's = .71 to .80), and high test-retest stability (ICC's = .92 to .95). Correlations of caregiver and rater global (but not patient only rating) CGI-CogS ratings with neurocognitive performance were in the moderate range (r's = -.27 to -.48), while most of the correlations with functional outcome were moderate to high (r's = -.41 to -.72). In fact, the CGI-CogS ratings were significantly more correlated with Social Functioning than were objective neurocognitive test scores (p = .02) and showed a trend in the same direction for predicting Instrumental Functioning (p = .06). We found moderate correlations between CGI-CogS global ratings and PANSS positive (r's = .36 to .49) and SANS negative symptoms (r = .41 to .61), but not with BPRS depression (r's = .11 to .13). Conclusions: An interview-based measure of cognition demonstrated high internal consistency, good inter-rater reliability, and high test-retest reliability. Caregiver ratings appear to add important clinical information over patient-only ratings. The CGI-CogS showed moderate validity with respect to neurocognitive performance and functional outcome, and correlations of CGI-CogS with functional outcomes were stronger than correlations of objective neurocognitive performance with functional outcomes. The CGI-CogS appears to offer a reliable and valid method for clinical rating of cognitive deficits and their impact on everyday functioning in schizophrenia. (C) 2007 Elsevier B.V. All rights reserved.