True findings about schizophrenia remain elusive; many findings are not replicated and conflicting results are common. Well-conducted systematic reviews have the ability to make robust, generalizable conclusions, with good meta-analyses potentially providing the closest estimate of the true effect size. In this paper, we undertake a systematic approach to synthesising the available evidence from well-conducted systematic reviews on schizophrenia.

Reviews were identified by searching Medline, EMBASE, CINAHL, Current Contents and PsycINFO. The decision to include or exclude reviews, data extraction and quality assessments were conducted in duplicate. Evidence was graded as high quality if reviews contained large samples and robust results; and as moderate quality if reviews contained imprecision, inconsistency, smaller samples or study designs that may be prone to bias.

High- and moderate-quality evidence shows that numerous psychosocial and biomedical treatments are effective. Patients have relatively poor cognitive functioning, and subtle, but diverse, structural brain alterations, altered electrophysiological functioning and sleep patterns, minor physical anomalies, neurological soft signs, and sensory alterations. There are markers of infection, inflammation or altered immunological parameters; and there is increased mortality from a range of causes. Risk for schizophrenia is increased with cannabis use, pregnancy and birth complications, prenatal exposure to Toxoplasma gondii, childhood central nervous system viral infections, childhood adversities, urbanicity and immigration (first and second generation), particularly in certain ethnic groups. Developmental motor delays and lower intelligence quotient in childhood and adolescence are apparent.

We conclude that while our knowledge of schizophrenia is very substantial, our understanding of it remains limited.