Background Few studies have examined the relationship between sociodemographic factors and a population-based measure of developmental delay in US children. We identify sociodemographic factors associated with unlikely, probable and possible developmental delay in preschool US children using nationally representative data.

Methods All children aged 18 months to 5 years in the 2007 National Survey of Children's Health were categorised into three groups based on the likelihood of developmental delay (unlikely delay, possible delay and probable delay) using a modified survey version of the Parents’ Evaluation of Developmental Status questionnaire. Bivariate and multivariate multinomial logistic regressions were used to assess relations between sociodemographic variables and risk of developmental delay.

Results Children had increased odds of probable delay (compared with unlikely delay) if they were older (adjusted OR (aOR)=1.41/additional year above the youngest age group (18 months–2 years), p<0.001), male (aOR=1.55, p<0.001), low birth weight (aOR=2.08, p<0.001), non-Hispanic black (aOR=1.50, p<0.01) or Hispanic in a non-English-speaking household (aOR=2.53, p<0.001) versus non-Hispanic white, had lower household income (aOR=1.33 for each decreasing category of poverty level, p<0.001), or received >10 h/week of care at another family's home (aOR=1.71, p<0.05). Only four characteristics (being older, male, low birth weight and Hispanic living in a non-English-speaking household) were associated with increased odds of possible delay compared with unlikely delay.

Conclusions Multiple factors, including demographic characteristics and indicators of social disadvantage, distinguish children with probable developmental delay from those unlikely to have developmental delay. Fewer factors identify children with possible delay.