A systematic meta-review grading the evidence for non-genetic risk factors and putative antecedents of schizophrenia

Abstract

Introduction

Identifying the relative strength of evidence associated with non-genetic risk factors and putative antecedents of schizophrenia will guide research and may inform the design of early detection and intervention strategies.

Aims

To present and quality assess current evidence for non-genetic risk factors and putative antecedents derived from well-conducted systematic reviews that report pooled data.

Method

Medline, Embase, CINAHL, Current Contents, and PsycINFO databases were searched systematically, and supplemented by hand searching. Review reporting quality was assessed using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist, review methodology was assessed using the Assessment of Multiple Systematic Reviews (AMSTAR) checklist, and evidence quality was assessed using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach.

Results

Twenty-four reviews met inclusion criteria. The risk factors with the highest quality evidence, reporting medium effect sizes, were advanced paternal age, obstetric complications, and cannabis use. The strongest evidence among the putative antecedents was identified for motor dysfunction and low IQ.

Conclusions

More research is required that applies sound methodological practices, taking into consideration specificity for schizophrenia and possible confounding factors, to robustly identify the non-genetic risk factors and putative antecedents of schizophrenia.

Introduction

As part of a recent series of articles in this journal titled “Just the Facts”, Tandon et al., 2008a, Tandon et al., 2008b reviewed existing evidence regarding the aetiology of schizophrenia according to three criteria: reproducibility, whether primary to schizophrenia, and durability of the finding over time. Among the most established “Facts” is evidence that genetic factors contribute ~ 80% of the liability for schizophrenia. Yet to be established, however, is the relative strength of evidence associated with a variety of non-genetic/environmental factors that act interactively or additively to the genetic liability. Non-genetic factors arising early in life may lay the ground for future emergence of the disorder, while events occurring later in life may contribute to symptom onset (Weinberger, 1995, Cannon et al., 2003). Putative antecedents of schizophrenia are largely genetically determined, but may also be influenced by environmental factors. They are present in childhood and early adolescence, and might reflect early, passive expression of the disorder or may actively modify risk within the aetiological pathway (Carter et al., 2002, Welham et al., 2009).

Alongside research invested in identifying genetic determinants of schizophrenia, robust identification of non-genetic factors and antecedents offers the prospect of early detection of individuals at risk for later disorder, which in turn may enable delivery of targeted interventions to prevent the disorder, delay its onset, or reduce its severity, particularly if a unique schizophrenia-risk profile during childhood and adolescence can be established (Laurens et al., 2007). This profile might vary depending on the degree of risk exposure, familial risk, trait related predispositions such as schizotypy, and the expression of antecedents that might wax and wane over time (Keshavan et al., 2011).

A sound methodological approach is required to evaluate findings objectively (Tandon et al., 2008a, Tandon et al., 2008b). Systematic reviews that incorporate meta-analyses or pooled data have the greatest power to detect the magnitude and consistency of effects, with increasing magnitude and consistency implying a stronger association with causality (Mann, 2003, Chan et al., 2009). The most reliable data is provided by prospective longitudinal investigations of population cohorts and children at elevated genetic risk for schizophrenia by virtue of a family history of illness. These data are usefully supplemented by ‘follow back’ studies that utilize prospectively collected data obtained prior to illness onset (Mann, 2003). Data from case–control studies using retrospective self-report measures may be prone to recall bias, and population level ecological, incidence, and prevalence studies provide only broad indicators of risk. All are informative, but they are subject to a range of confounding factors, so the causal inference is not absolute (Rutter, 2009).

This meta-review aims to synthesise and quality assess the current evidence for non-genetic risk factors and putative antecedents derived from well-conducted systematic reviews incorporating meta-analyses or other data pooling techniques. The quality assessment uses objective techniques to grade study design, review reporting and methodology, and strength of evidence to identify non-genetic factors and antecedents with the greatest strength so that these may be prioritized for further study and utilized to inform the design of novel early detection and targeted intervention strategies.