Global Autism Spectrum Disorder Prevalence Estimates and Associated Covariates: A Systematic Review and...

This review aimed to estimate the global prevalence of autism spectrum disorder (ASD) and identify methodological and contextual covariates influencing prevalence estimates. Despite increasing research, the true global burden of ASD remains unclear due to substantial variability in reported prevalence across studies. The specific methodological and contextual factors driving this heterogeneity have not been fully quantified. Electronic databases, including PubMed, Scopus, Web of Science, Embase, and Google Scholar, were systematically searched from 2004 to 2025. All human population-based studies were included, irrespective of region or diagnostic framework. Nineteen studies, comprising approximately 20.6 million participants, met the inclusion criteria and reported ASD prevalence. To estimate pooled prevalence, a random-effects model (REML) was used, and Bayesian hierarchical modeling was conducted to provide posterior estimates. Publication bias and sensitivity analyses were conducted, followed by meta-regression to account for covariates such as diagnostic framework, study setting, and region. The frequentist pooled prevalence of ASD was 0.8% (95% CI: 0.4%-1.7%), with substantial heterogeneity (I² ≈ 100%) and a wide prediction interval (0.03%-17.3%). Bayesian hierarchical modeling produced a posterior mean prevalence of 1.55% (95% CrI: 0.75%-4.1%), and a Bayesian sensitivity analysis excluding an extreme outlier study yielded a posterior mean prevalence of 0.9% (95% CrI: 0.6%-1.8%) with substantial heterogeneity (τ = 1.14, I² = 99.98%) and a prediction interval of 0.09%-9.5%, indicating strong consistency between frequentist and Bayesian estimates. Regional prevalence varied: North America (1.9%), Africa (6.3%) (wide confidence intervals reflecting limited data), Europe (0.4%), Latin America (0.2%), and the Middle East (0.6%). Meta-regression analysis demonstrated that ASD prevalence was influenced by diagnostic framework, study setting, and region (R² ≈ 0.78), and the combined impact of all covariates explained a substantial proportion of the variance in effect sizes. Our updated meta-analysis indicated that global ASD prevalence varies widely across regions, largely due to methodological and contextual differences rather than demographics alone. Standardized diagnostic practices, enhanced surveillance, and targeted investment in underrepresented regions are critical for improving the accuracy and comparability of prevalence estimates. This updated synthesis provides a more robust and context-sensitive estimate of global ASD prevalence and supports the interpretation of variability observed in prior studies.