28178
Infection and Fever during Pregnancy and Risk for Autism: A Systematic Review and Meta-Analysis
A multifactorial model for ASD causation is increasingly recognized. Epidemiological family and twin studies have firmly established a genetic component underlying ASD with overall heritability estimates of approximately 50%. At the same time, these heritability estimates suggest the importance of non-genetic or environmental factors. Of the possible environmental risk factors for ASD, none has been studied as extensively as prenatal maternal infection, the subject of this review.
Objectives:
The objective of this study is to determine the association between maternal infection and fever and the risk of ASD through a meta-analysis of the extant literature.
Methods:
We conducted a comprehensive search of the PubMed and Google Scholar databases for peer-reviewed case-control studies and cohort studies written in English that examined the association between maternal infection during pregnancy and/or fever and/or antibiotics and the risk of ASD. The reference lists of retrieved articles were hand-searched for additional relevant articles. A total of 35 papers were ultimately included in the meta-analysis. From each paper, we extracted information pertaining to study design, the population under study, total number of subjects in each group, information regarding the type and timing of infection exposure, diagnostic criteria, outcome measures, and statistical adjustments. ORs were combined across studies using Comprehensive Meta-Analysis software (Borenstein et al. 2009). When individual articles examined multiple types of exposure, we collapsed both the effects to obtain study-level effects sizes (any infection or fever), and examined the exposure types separately in order to determine if there is heterogeneity of effect across infectious agents or across infections sites.
Results:
Study-level analyses revealed a modest, but significant effect of exposure to infection/fever on autism risk [OR=1.20;95%CI,1.07-1.35]. There is little or no heterogeneity by either infectious agent or infection site. Some heterogeneity exists across trimester of exposure, with first trimester exposures have slightly weaker effects than those occurring later, but these differences are not statistically significant. In rare instances when investigators considered variation in exposure severity (e.g. length of maternal exposure to fever) severity was often strongly related to effect size.
Conclusions:
Maternal exposure to infection confers a modest, but statistically reliable, average increase in risk of autism in offspring. The modest size of the increase risk suggests that the impact of maternal infection may need to be understood in terms of its interaction with other risk factors (e.g. genetic vulnerability). Although severity of exposure is under-studied, the larger odds ratios associated with lengthy bouts of fever, for example, suggests that these more severe exposures require fewer additional vulnerabilities to result in ASD. The absence of heterogeneity across infectious agent and site is consistent with the idea of a common mediator (e.g. markers of inflammation). The absence of heterogeneity across trimester could be the result of measurement error, or due to the relevant neurodevelopmental processes straddling these divisions.