Blood glucose and schizophrenia: a systematic review of prospective randomized clinical trials.

Abstract

Most of the data evaluating the potential relationship between diabetes, schizophrenia, and anti-psychotics currently derive from retrospective analysis. Relevant confounders of such data include screening and selection bias. Prospective data collected from randomized controlled trials may reduce such biases. As no single trial has glucose comparisons as a primary endpoint, we undertook a systematic review of available data. Embase, HealthStar, MEDLINE, Pre-MEDLINE, and PsycINFO databases were searched online for relevant articles. Abstracts from major congresses held between January 2000 and April 2006 were included. Search terms included all currently available antipsychotics: olanzapine, risperidone, clozapine, quetiapine, ziprasidone, aripiprazole, haloperidol, chlorpromazine, and zotepine. Prospective clinical trials involving schizophrenia patients with no stated previous glucose abnormalities randomly assigned to cohorts receiving active or placebo comparator antipsychotic medications were included with no restrictions on study length. 16 studies were from peer-reviewed publications, 4 were from posters at major congresses, and 2 were available only on Internet-based sites. Glucose parameters reported included fasting and random glucose and glycosylated hemoglobin. Data reported included mean changes and categorical reports of abnormal levels. Data were available in 6329 patients from 22 trials. The most common comparator agents were aripiprazole and olanzapine in 4 studies including 1432 patients. 14 studies reported fasting and 9 studies reported nonfasting data. 15 studies were a minimum of 5 months, with 8 studies of at least 1 year's duration. No consistent significant glucose differences were found between any comparator antipsychotics or placebo in any trial. In contrast to some of the retrospective data, an analysis of prospective data from randomized clinical trials showed no consistent significant differences in the incidence of treatment-emergent glucose abnormalities in patients treated with antipsychotics. The reduction in both screening and selection biases may be relevant. Although one third of the studies had at least 1 year's duration, the data are not sufficient to reach conclusions regarding patients receiving longer-term treatment with atypical antipsychotics.