Please, Don’t Shoot the Meta-analysis: A Response to “A Commentary to Toddler Screening for Autism Spectrum Disorder: A Meta-analysis of Diagnostic Accuracy by Sánchez-Garcia et al. 2019”

Abstract

[EN] In their letter to the editor, Øien et al. (2019) highlight concerns about conducting a meta-analysis on screening instruments for ASD. They focus on lack of information and heterogeneity among the individual studies pooled in the research by Sánchez-García et al. (2019). The authors do not question the experimental method; even stating that it “provides us with important knowledge of how screening instruments perform across various studies.” Of course, clinical and methodological heterogeneity could cause statistical heterogeneity leading to inaccurate conclusions in a meta-analysis (Higgins and Altman 2008). Although this should not preclude conducting meta-analytic investigation, since understanding the causes of heterogeneity increases its scientific value and the clinical significance of the results (Thompson 1994). Therefore, it is not a threat or concern to synthesize the available evidence (Lijmer et al. 2002). Having said that, the methodological limitations among studies included in a meta-analysis is not specific to ASD screening; yet, there is a large literature about the viability and importance of diagnostic or screening accuracy meta-analyses. Given empirical support in other disorders, it is logical to apply them to a meta-analysis in screening instruments for ASD. There are three specific points that warrant further consideration.