Knowledge and Misinformation Regarding Autism Risk Factors in a Survey of US Adults

Importance: With the rise of misinformation about autism risk, including recent statements from US government officials, the extent to which the public understands environmental autism risk factors is not known. Objective: To characterize beliefs about autism risk factors among US adults and identify characteristics influencing these beliefs. Design, Setting, and Participants: This cross-sectional survey study used data from an internet-based survey of adults aged 18 years or older residing in all 50 US states and the District of Columbia and was administered between December 18, 2025, and January 27, 2026. Multiple regression models assessed sociodemographic and other individual characteristics, including political affiliation and trust in institutions, for association with endorsement of misinformed beliefs. Main Outcomes and Measures: The primary outcome was endorsement of individual risk factors, scored on a 5-point Likert scale from definitely false to definitely true. Results: Of the 30 388 adults who completed the survey, 100 did not provide an answer to the childhood vaccination item, leaving 30 288 in the analysis. Among the 30 288 participants (mean [SD] age, 46.6 [15.9] years), 13 702 (45.2%) identified as men, 16 393 (54.1%) as women, and 193 (0.6%) as nonbinary. The most commonly endorsed cause was genetic factors (20 380 of 30 254; unweighted, 67.4%; weighted, 66.7%), an evidence-based risk factor. However, 8370 of 30 288 (unweighted, 27.6%; weighted, 26.6%) endorsed childhood vaccinations and 7140 of 30 211 (unweighted, 23.6%; weighted, 23.1%) endorsed acetaminophen in pregnancy as autism risk factors. In regression models, endorsement of vaccine-autism risk beliefs was greatest among adults aged 35 to 44 years (vs 18 to 24 years, adjusted odds ratio [aOR], 1.64; 95% CI, 1.44-1.86) and lowest among those aged 65 years and older (aOR, 0.64; 95% CI, 0.55-0.74). Endorsement was also associated with men (aOR, 1.26; 95% CI, 1.19-1.34), African American or Black race (compared with White race, aOR, 1.25; 95% CI, 1.15-1.35), Hispanic ethnicity (aOR, 1.13; 95% CI, 1.03-1.24), and strength of Republican political affiliation (strong Republican vs strong Democrat, aOR, 3.83; 95% CI, 3.47-4.24). More frequent social media use was associated with lower odds of endorsing misinformation (multiple times daily vs less than weekly, aOR, 0.72; 95% CI, 0.66-0.78). Conclusions and Relevance: Results of this survey study suggest that non-evidence-based beliefs about autism risk were prevalent among US adults and were strongly associated with sociodemographic differences and political affiliation. Limiting public health impact of such non-evidence-based information will likely require targeted interventions that recognize the heterogeneity of such beliefs.