Fig. 1.

PRISMA flow diagram. After screening by title and abstract, 168 records remained. Duplicates (same studies in several databases) were then excluded (n = 71 articles). Ninety-seven articles remained after the duplicates had been removed. When articles not available in full text were removed, 85 articles remained. The full-text papers were screened according to the eligibility criteria, and 65 papers were excluded leaving 20 studies to be included in the review. Eleven of the remaining 20 studies were based on data from the National Survey of Children's Health (NSCH). Only one study per NSCH survey year was included in the meta-analysis, and the selected study should comprise the broadest age span and the largest number of individuals with autism. Six NSCH studies were therefore excluded (1 from NSCH year 2004, 3 from NSCH year 2012, and 2 from NSCH year 2017), leaving 14 studies that were included in the meta-analysis.