A systemic review of maternal infections and schizophrenia

Background Findings for the association of prenatal maternal infections and schizophrenia are inconsistent. Whether specific maternal infections may contribute to central nervous system dysfunction, like schizophrenia remains unknown. Thus, a systematic review is conducted to clarify the role of infections on the pathogenesis of schizophrenia. Methods A manual and electronic search of the literature for prenatal maternal infection at the individual level and the diagnosis of schizophrenia in offspring was conducted. All the included studies are searched from PubMed and EMBASE databases and generally based on population level datasets with cohort or nested case-control studies designs. We only focus on three kinds of maternal infections in the review (i.e. influenza pandemic, rubella and Toxoplasmosis), other infections such as herpes simples, varicella‐zoster virus and polio were excluded. Positive family history of psychotic illness was excluded as genetic disposition is established as a strong risk factor of developing schizophrenia in offspring. Results Fifteen studies were included in this systemic review. Nine (60%) of these studies reported a positive association between maternal infections and the development of schizophrenia in the later lives of their offspring, while six (40%) of these studies indicated that there is no significant difference among cases of schizophrenia in maternal infections exposure and the control subjects or no association between schizophrenia in offspring and the infections during pregnancy as the exposure. Among all the studies, thirteen cohort studies and two nested‐case studies were examined. Eight (61.5%) showed a positive association, meanwhile five (38.5%) showed a null association. Two nested‐case control studies both suggested that maternal infection exposure is a risk factor for schizophrenia. Conclusion Findings for the association of prenatal maternal infections and schizophrenia are not equal to the evidence for causation. Due to the absence of external validity due to small and non‐representative samples, selection bias was suspected even with a significant result. Additionally, multiple exposures were hard to testify during pregnancy. Even we adjusted for the measured confounders, residual confounding such as genetic disposition and socioeconomic status, as well as unmeasured confounding are concerns. Quasi-experimental methods may help improve causal inference such as the use of instrumental variable analysis in future studies. Preventive strategies such as immunizations need to be deliberated carefully to weight the benefit and the potential impact.published_or_final_versionPublic HealthMasterMaster of Public Healt