Examining Sex Differences in Item Endorsement on the Modified Checklist for Autism in Toddlers (M-CHAT)

Background: Epidemiological studies of Autism Spectrum Disorders (ASDs) indicate 1% prevalence and a 4:1 male-to-female ratio, and it is expected that early autism screening tools would detect boys and girls at a similar rate. Despite this, our previous findings demonstrate that girls show higher false positive rates on the Modified Checklist for Autism in Toddlers (M-CHAT; Robins et al., 1999).

Objectives: To examine the characteristics of items endorsed differentially for boys and girls on the Modified Checklist for Autism in Toddlers to better understand the differential predictive value of this tool based on sex.

Methods: The sample included 9601 males and 9140 females (mean age=20.35 months, SD=3.10) screened for ASDs using the M-CHAT, a parent-report questionnaire, at toddler’s pediatric well-visits. Parents of children who screened positive on the M-CHAT were offered the M-CHAT Follow-up Interview (FUI). Those who screened positive on the FUI were offered a clinical evaluation.

Results: For the total sample, mean total scores on the M-CHAT differed based on sex, M_male=1.00, SD=1.70, M_female=.80, SD=1.31, t(18739)=9.11, p<.001 d=.13 Of the subsample who completed the FUI, mean total scores differed based on sex, M_male= 1.78, SD=2.87, M_female=1.15, SD=2.16, t(1277)=4.23, p<.001 , d=.25. Chi-square analyses indicated that 11 of 23 items on the M-CHAT were endorsed more often in boys than girls (p’s<.001) and no items were endorsed more in girls than boys. On the FUI, only one out of 23 items was endorsed more often in boys than girls (pointing to show interest; p=.001). A discriminant function analysis yielded nine items in boys and eleven items in girls that best predicted ASD. Eight of the critical items overlapped (interest in peers, brings objects to show, response to name, point to show interest, wondered if deaf, follows point, pretend play, response to smile), but point to request was a better predictor of a diagnosis in females and understands what others say was a better predictor in males.

Conclusions: Overall, parents endorse more items for boys than girls on both the M-CHAT and FUI. However, only half of the M-CHAT items are significantly endorsed more often in boys than in girls, and this decreases to one item at the FUI level. This suggests that half of the M-CHAT’s items may be contributing to reduced PPV in girls compared to boys. The critical items for girls and boys are largely overlapping, but there are some differences in item content. Future research will examine sex-specific scoring algorithms to maximize sensitivity and specificity.