Considerations Regarding Autism Spectrum Disorders and Cardiometabolic Diseases

Abstract

To the Editor In their systematic review and meta-analysis, Dhanasekara et al 1 evaluated the association between autism spectrum disorders and cardiometabolic diseases. The findings suggest that autism was associated with an increased risk of diabetes, dyslipidemia, and atherosclerotic heart disease. Children with autism may have a higher risk of developing diabetes and hypertension. We commend the authors for their effort to synthesize the conflicting findings in the literature to guide the prevention of cardiometabolic diseases. However, we have 2 comments on this article.First, a major concern is the criteria for data extraction from the studies that report cardiometabolic results (ie, diabetes and hypertension) only in adults or children vs those reporting results in both adults and children is unclear. For example, we observed that 2 studies by Chen et al 2 and Flygare Wallén et al 3 reported effect estimates of relative risk (or these were calculated by Dhanasekara et al 1 ) for adults and children separately, but these 2 studies are categorized into different subgroups (eFigure 2 in Supplement 1 1 ). Considering this issue, we suggest that studies that included adults and children can be included in both subgroups if relevant outcomes for each age group are reported separately. 4 To be more informative when evaluating the difference by age groups, we propose conducting a sensitivity analysis from only those cohort studies that report results for both adults and children. It may help decrease between-study variations for adults and children.Second, it is challenging to analyze this topic due to confounding factors. Though the authors considered several sociodemographic variables in meta-regression analyses, residual confounding, a typical limitation of observational studies, might still exist. We suggest this issue might be worth mentioning in the Discussion. In addition, cardiometabolic outcomes usually take a long time (even decades) to develop. 2,5 To strengthen the findings, it would be instructive to report the length of follow-up in each study identified; a sensitivity analysis by different follow-up periods (such as 5 or 10 years) could then be performed to examine the robustness of the associations.