Objective  To conduct exploratory factor analysis on the Checklist for Autism in Toddler (CHAT-23) and to carry out autism screening among small children with CHAT-23.

  Methods  We recruited 2 358 children aged 16 – 36 months and having routine physical examination at 22 community healthcare centers in two randomly selected central urban districts of Tianjin city and carried out autism screening with CHAT-23 between February and October 2013. Principal component analysis and Promax rotation method were used in exploratory factor analysis on CHAT-23.

  Results  The exploratory factor analysis revealed 4 common factors in CHAT-23, namely children's response, interaction with parents/strangers, pointing movement, and interactive game, which explained 50.87% of the total variation of the scale. Of the 2 358 children screened, 121 were positive for autism, with a screening positive rate of 5.13%. Among the 1 238 children with complete demographic information, there were significant differences between the autism positive and negative children in sex ratio (χ² = 4.77, P = 0.03), maternal history of threatened abortion (χ² = 4.02, P = 0.05), and maternal history of taking medicine during pregnancy (χ² = 6.17, P = 0.01). Regression analysis reveal that being a boy (odds ratio OR = 1.92, 95% confidence interval 95% CI: 1.08 – 3.42) and with history of taking medicine during pregnancy (OR = 2.66, 95% CI: 1.01 – 6.70) were risk factors of autism positive.

  Conclusion  CHAT-23 is of good structural validity and could be used as a screening tool in primary health care institutions.