Prenatal and perinatal insults are implicated in the aetiopathogenesis of psychotic disorders but the consistency and magnitude of their associations with psychosis have not been updated for nearly two decades. The aim of this systematic review and meta-analysis was to provide a comprehensive and up-to-date synthesis of the evidence on the association between prenatal or perinatal risk and protective factors and psychotic disorders.
In this systematic review and meta-analysis, we searched the Web of Science database for articles published up to July 20, 2019. We identified cohort and case-control studies examining the association (odds ratio [OR]) between prenatal and perinatal factors and any International Classification of Diseases (ICD) or Diagnostic and Statistical Manual of Mental Disorders (DSM) non-organic psychotic disorder with a healthy comparison group. Other inclusion criteria were enough data available to do the analyses, and non-overlapping datasets. We excluded reviews, meta-analyses, abstracts or conference proceedings, and articles with overlapping datasets. Data were extracted according to EQUATOR and PRISMA guidelines. Extracted variables included first author, publication year, study type, sample size, type of psychotic diagnosis (non-affective psychoses or schizophrenia-spectrum disorders, affective psychoses) and diagnostic instrument (DSM or ICD and version), the risk or protective factor, and measure of association (primary outcome). We did random-effects pairwise meta-analyses, Q statistics, I2 index, sensitivity analyses, meta-regressions, and assessed study quality and publication bias. The study protocol was registered at PROSPERO, CRD42017079261.
152 studies relating to 98 risk or protective factors were eligible for analysis. Significant risk factors were: maternal age younger than 20 years (OR 1·17) and 30–34 years (OR 1·05); paternal age younger than 20 years (OR 1·31) and older than 35 years (OR 1·28); any maternal (OR 4·60) or paternal (OR 2·73) psychopathology; maternal psychosis (OR 7·61) and affective disorder (OR 2·26); three or more pregnancies (OR 1·30); herpes simplex 2 (OR 1·35); maternal infections not otherwise specified (NOS; OR 1·27); suboptimal number of antenatal visits (OR 1·83); winter (OR 1·05) and winter to spring (OR 1·05) season of birth in the northern hemisphere; maternal stress NOS (OR 2·40); famine (OR 1·61); any famine or nutritional deficits in pregnancy (OR 1·40); maternal hypertension (OR 1·40); hypoxia (OR 1·63); ruptured (OR 1·86) and premature rupture (OR 2·29) of membranes; polyhydramnios (OR 3·05); definite obstetric complications NOS (OR 1·83); birthweights of less than 2000 g (OR 1·84), less than 2500 g (OR 1·53), or 2500–2999 g (OR 1·23); birth length less than 49 cm (OR 1·17); small for gestational age (OR 1·40); premature birth (OR 1·35), and congenital malformations (OR 2·35). Significant protective factors were maternal ages 20–24 years (OR 0·93) and 25–29 years (OR 0·92), nulliparity (OR 0·91), and birthweights 3500–3999 g (OR 0·90) or more than 4000 g (OR 0·86). The results were corrected for publication biases; sensitivity and meta-regression analyses confirmed the robustness of these findings for most factors.
Several prenatal and perinatal factors are associated with the later onset of psychosis. The updated knowledge emerging from this study could refine understanding of psychosis pathogenesis, enhance multivariable risk prediction, and inform preventive strategies.